Probabilistic
Seismic Hazard Maps of Alaska
by Robert L. Wesson1,
Arthur D. Frankel1, Charles S. Mueller1, and Stephen
C. Harmsen1
Open-File
Report 99-36
1999
This report is preliminary and has not been reviewed
for conformity with the U.S. Geological Survey editorial standards or with
the North American Stratigraphic Code. Any
use of trade, firm or product names is for descriptive purposes only and does
not imply endorsement by the U.S. Government.
U.S. DEPARTMENT OF THE INTERIOR
U.S. GEOLOGICAL SURVEY
1 Denver, Colorado
Abstract
Probabilistic
seismic hazard maps have been prepared for Alaska portraying ground motion
values (peak ground acceleration and spectral amplitude at periods of 0.2,
0.3 and 1.0 seconds) at probabilities of exceedance of 2% and 10% in 50 years. Preparation
of these maps followed the same general strategy as that followed for the
U.S.G.S. seismic hazard maps of the contiguous United States, combining hazard
derived from spatially-smoothed historic seismicity with hazard from fault-specific
sources. Preparation of the Alaska
maps presented particular challenges in characterizing the hazard from the
Alaska-Aleutian megathrust. In the maps of the contiguous United States
the rate of seismicity for recognized active faults was determined from slip
rates estimated from geologic data. This
approach is not appropriate for the megathrust because it has been demonstrated
that a significant fraction of the subduction occurs aseismically. The characteristic earthquake hypothesis, based
on recurrence rates determined from geologic data, is appealing for the portion
of the megathrust that ruptured in the 1964 Alaskan earthquake, but is shown
to be inappropriate for the western portion of the megathrust by the recent
large earthquakes in the region which did not follow the characteristic model.
Consequently the hazard from the western portion was estimated based
on a truncated Gutenberg and Richter model derived from historic seismicity,
and the hazard for the 1964 zone was estimated from a combination of a Gutenberg
and Richter model derived from historic seismicity and the characteristic
earthquake hypothesis with recurrence rates estimated from geologic data.
Owing to geologic complexity and limited data, hazard models of the
easternmost portion of the megathrust in the vicinity of Yakataga are the
least satisfactorily constrained. Hazard
is estimated for the recognized crustal faults of the Denali, Fairweather-Queen
Charlotte and Castle Mountain fault systems based on available geologic slip
rates. Hazard from other sources is estimated from
spatially smoothed historic seismicity. Disaggregations
of the hazard for Anchorage, Fairbanks and Juneau reveal the dominant sources
of the hazard at each location.
Introduction
Alaska is the most seismically active state in the
United States, and in 1964 the site of one of the largest earthquakes since
the beginning of instrumental recording.
Although the current population of the region is small by comparison
with, say California, the consequences of a large earthquake in the region
could be much greater now than at the time of the 1964 Alaskan earthquake. The probabilistic seismic maps we have prepared
are intended to extend those prepared by Frankel et al. (1996) for the 48
contiguous states, and with soon-to-be-published maps for Hawaii. Our methodology follows the basic approach
of Cornell (1968). These maps are
intended to summarize the available quantitative information about seismic
hazard from geologic and geophysical sources.
Full color maps at a scale of 1:7,500,000 are available as U.S. Geological
Survey Miscellaneous Investigations Series I-2679 (Wesson et al., 1999).
The process of preparing these maps included a workshop held in Anchorage
in the fall of 1996 attended by many scientists and engineers involved in
aspects of earthquake research and structural engineering practice in Alaska.
Preliminary calculations of hazard were presented for discussion and
a number of recommendations were made that affected the subsequent preparation
of the map. Draft maps were posted on the World Wide Web and circulated for comment in the fall of 1997.
The maps presented here have benefited greatly from both the original
workshop in 1996 and from the review comments received.
The strategy for preparing these maps is similar
to that for recently prepared seismic hazard maps of the contiguous United
States (Frankel et al., 1996). The
maps presented here include maps for peak ground acceleration and 1.0, 3.3
and 5.0 Hz spectral acceleration at probabilities of exceedance of 10% in
50 years (annual probability of 0.0021) and 2% in 50 years (annual probability
of 0.000404).
The historical instrumental seismicity of Alaska
and the Aleutians for earthquakes greater than or equal to magnitude 6 is
shown in Figure 1.
The preparation of the earthquake catalogs used for the analyses in
this report (and shown in Figure 1) is
discussed in a companion report, Mueller et al., 1998).
Clearly the majority of the seismicity in the region is associated
with the Alaska-Aleutian megathrust fault extending eastward along the Aleutian
arc into south central Alaska. The northwestward-moving Pacific plate is subducted
along this megathrust beneath the North American Plate giving rise to the
Aleutian trench and islands. Additional
significant seismicity occurs along the northwestward-striking system of right-lateral
strike-slip faults extending southeastward through and offshore of the panhandle
of southeast Alaska. This system of faults forms the northeast boundary
of the Pacific plate. Additional seismicity
occurs in central Alaska.
The estimated rupture zones of the largest earthquakes
in this century are shown in Figure 2 (Plafker
et al., 1993). During this century
virtually the entire plate boundary from the westernmost Aleutian Islands
to the Queen Charlotte Islands off British Columbia has ruptured in large
to great earthquakes. The only exceptions are areas near the Komandorsky
Islands, near the Shumagin Islands, and near Cape Yakataga (Sykes, 1971; Davies et al., 1981). Near the Komandorsky Islands, historical records
of large earthquakes in 1849 and 1858 at the extreme western end of the arc
have been judged as insufficient to conclude that plate-margin-rupturing earthquakes
have occurred there (Sykes et al., 1981; Taber et al., 1991). At this location subduction is occurring at
a highly oblique angle, and it has been argued that the recurrence properties
of large earthquakes here may differ significantly from those else along the
arc. Indeed, Cormier (1975) has argued
that the region may be incapable of supporting a great earthquake. In the vicinity of the Shumagin Islands, that
is, in the region between the 1957 and 1938 earthquakes, it has been argued that no great earthquake
has occurred in this century. Similarly,
the vicinity of Cape Yakataga has experienced no great earthquakes in this
century. These two regions have been
identified as "seismic gaps," that is, the potential sites of future
large earthquakes (Davies et al., 1981; Sykes, 1971, Lahr et al., 1980).
Characterization
of Seismic Sources
The seismic potential of Alaska is captured through
consideration of earthquake sources that can be explicitly identified, including
the Alaska-Aleutian megathrust and active crustal faults with known slip rates;
and earthquake sources that are characterized by spatially smoothed historical
seismicity, including shallow earthquakes from sources not included above,
and deeper earthquakes (focal depths of 50 to 120 km). Hazard calculated from all these sources was
combined as shown in Figure 3.
Spatially-smoothed seismicity was used to estimate the hazard from
shallow crustal earthquakes in the magnitude range 5.0 to 7.3 based on a Gutenberg-Richter
model. For the deeper earthquakes, hazard was estimated
for the magnitude range 5.0 to 7.0 based on a Gutenberg-Richter model with
the parameters estimated from the spatially-smoothed seismicity in the two
depth intervals, 50 to 80 km, and 80 to 120 km.
Alaska-Aleutian
megathrust
The Alaska-Aleutian megathrust has been responsible
for several of the largest earthquakes known in instrumental seismology, including
the 1964 Prince William Sound (Mw 9.2) and 1957 Aleutian (Mw 9.1) earthquakes
(Figure 2).
For purposes of this analysis two different segmentation models of
the megathrust with corresponding recurrence assumptions have been considered.
Segmentation models
In the first
segmentation model (Model I), the megathrust has been divided into three parts
as shown in Figure 4.
The Western Zone includes the reach of the megathrust along the western
and central Aleutians from about 170.2° E to 161.7° W.
The Eastern Zone extends eastward and includes the reach along the
eastern Aleutians, the Alaska Peninsula and Prince William Sound, from about
161.7° W to 144.2° W. Finally, the
Yakataga Zone, extends further to the east, across the Copper River delta to Yakutat, that is, to about 139.5° W. In view of the uncertainties in the earthquake
potential in the region of the Komandorsky Islands, no segment of the megathrust
was explicitly modeled west of 170.2° E. These boundaries are interpreted to be the
limits of the possible rupture surfaces of significant earthquakes associated
with the megathrust.
In the second segmentation model of the megathrust
(Model II), shown in Figure 5, four segments
are considered. This model is similar to the first, in that
the Western Zone in Model I generally corresponds to Zone A in Model II, and
the Yakataga Zones in the two models are identical. The primary differences between Models I and II arise in the region
between the eastern Aleutians and Prince William Sound. In Model II, the Eastern Zone of the Model I is divided into Zones B and C.
The portion of the Eastern Zone that ruptured in the 1964 earthquake
is identified as Zone C, and the remainder, including the portions of the
Eastern Zone that ruptured in the 1938 earthquake and that portion between
the 1938 and 1957 earthquake ruptures are included in Zone B.
It seems reasonable to conclude that an earthquake
of magnitude 9.2 is the maximum to be expected from the megathrust, but it
is not clear whether the potential for an earthquake of this magnitude extends
throughout the Eastern Zone of Model I (that is as far west as the limit of
the continental crust near 161.7° W), or whether this potential is limited
the rupture area of the 1964 earthquake.
The occurrence of the 1938 earthquake (Mw 8.2) off the lower Alaskan Peninsula argues that the behavior of
this portion of the region may not be characteristic with a magnitude of 9.2.
Model II is intended to that this possibility into account, confining
the characteristic magnitude 9.2 to the 1964 aftershock zone.
In Model I the boundary between the Western and Eastern
zones, and in Model II the boundary between Zones A and B, is taken as the
approximate limit of continental crust north of the Aleutian Arc. This segment
boundary was suggested by several participants at our workshop. Specifically, we set the boundary at the eastern
edge of the 1946 earthquake. The selection of this boundary is based on the
assumption that it is extremely unlikely that a large earthquake could rupture
through this region. A consequence
of prohibiting rupture through a boundary is that the calculated hazard shows
a saddle in the region of the boundary. The saddle occurs because the hazard is calculated from a sequence
of floating rupture zones that are offset incrementally along the megathrust.
Because no rupture zones are allowed to cross the segment boundary,
a site at the end of the segment is immediately adjacent to only one floating
rupture, and is at increasing distances from all other floating ruptures as
they are offset along the megathrust. In
contrast, higher hazard is calculated at the center of a segment where a site
is immediately adjacent to many floating rupture zones.
Reviewers objected to the presence of the saddle, but were unwilling
to abandon the concept of a segment boundary in this region. To resolve this incompatibility, in Model I
the Western and Eastern Zones were allowed to overlap by 200 km. The resulting hazard is generally constant
along the strike of the megathrust through this region. The eastern limit of the Eastern Zone is taken
to be the approximate eastern limit of the aftershocks of the 1964 earthquake
(and of the well-defined Alaska-Aleutian Benioff zone). The western boundary
of the Western Zone is taken as the western limit of the 1965 Rat Islands
earthquake rupture zone (approximately 170° E).
Although a variety of more detailed segmentation
models have been proposed for the megathrust zone, large earthquakes, particularly
in the western and central Aleutians, notably the 1986 earthquake (Mw 8.0)
near Adak Island, have tended to occur without particular regard for the proposed
boundaries.
The southern or updip boundaries of the Western and
Eastern Zones in Model I and Zones A, B and C in Model II are defined by the
so called "seismic front" that is, the presumed updip limit of that
part of the megathrust capable of producing a significant earthquake. The seismic front is generally defined by the
southern limit of well-recorded seismicity (Engdahl, written communication,
1997) and by the break in slope at the lower edge of the shelf on the northern
side of the Aleutian Trench. (This
break approximately follows the 400 m depth contour, c.f. Plafker et al.,
1993). The updip boundary is taken
to be at a depth of 20 km.
The northern or downdip boundaries of the Western
and Eastern Zones, except in the area of the Alaskan Peninsula and Cook Inlet,
is taken as the 50 km depth contour of earthquakes in the Benioff zone (Plafker
et al., 1993). Boyd et al. (1995)
observed that standard earthquake locations in the central and western Aleutians
are biased as much as several 10's of km to the north, owing to the influence
of early arrivals at seismograph stations in Europe to which wave propagation
is along anomalously high-velocity paths down the subducting slab. In considering the appropriate northern boundaries
of the zones, hypocenters of earthquakes relocated by Engdahl (written communication,
1997) were compared with the map and depth contours of Plafker et al. (1993)
and found to be in good agreement. In the eastern Alaskan Peninsula and Cook Inlet
regions, the aftershocks of the 1964
earthquake did not extend as far downdip (or north) as the 50 km contour,
and the boundary was taken as the approximate northern limit of the aftershocks
of the 1964 earthquake (Plafker et al., 1993). In this region the boundary was assumed to lie at a depth 40 km.
Between about 155.0° W and 159.2° W the depth to the boundary increases
smoothly westward from 40 to 50 km. The assumption about the location of the downdip
or northern limit of the megathrust in
the Cook Inlet area is important because it significantly affects the estimation
of hazard in the area of Anchorage.
Although the Yakataga segment is clearly the location
of significant north-south convergence and the site of very large earthquakes
(e.g. 1899, 1979) the details of the
faulting are poorly understood. Several
east-trending, north-dipping thrust faults are inferred to exist beneath the
heavily glacier-covered region. As
a proxy for a more detailed understanding, a flat fault surface (that is,
with a 0° dip) at a depth of 15 km was assumed, extending from 59.1° to 61.0°
N and from 139.5° to 145.4° W.
Recurrence Assumptions
Although the rate of convergence across the Alaska-Aleutian
megathrust is relatively well known, the fraction of the convergence that
is accommodated by large earthquakes is significantly less than one.
This fraction varies with position along megathrust zone, but is poorly
known. It appears to range from 10% or less to near 100% (c.f. Pacheco
et al., 1993). Thus, estimates of
the rate of large earthquakes based on the known convergence rate alone are
unrealistically large, well above the observed rate of large earthquakes over
the last century.
Other than the plate convergence rate, the only data
available to estimate recurrence along the entire megathrust is the instrumental
seismicity catalog. At present geologic
data for recurrence exists only for the 1964 zone. Plafker and Rubin (1994) estimate that seven
or eight events with displacements similar to 1964 are reflected in the stratigraphic
sequence in the Copper River delta in the ~5600 years preceding 1964.
These data suggest a recurrence time for earthquakes of magnitude 9.2
of 700 to 800 years. The occurrence of the 1938 earthquake (Mw 8.2) off the lower Alaskan
Peninsula indicates that very large earthquakes with magnitudes less than
9.2 occur as well. Therefore, one
must ask the question, "What is the largest earthquake that might have
occurred within the 1964 zone, but would not be reflected in the stratigraphy
of the Copper River delta?" Our
estimate is that an earthquake with a magnitude as large as 8 could occur
without causing sufficient vertical displacement in the region of the Copper
River delta to be reflected in the stratigraphy.
In view of the limited geologic data, recurrence
assumptions for most of the megathrust are based on instrumental seismic data.
Hazard was estimated assuming a Gutenberg-Richter recurrence model
for magnitude 7.0 to 9.2 in the Western Zone of Model I and Zone A of Model
II. Similarly, a Gutenberg-Richter
recurrence model was assumed for magnitudes 7.0 to 8.5 in Zone B of Model
II, for magnitudes of 7.0 to 8 in the Eastern Zone of Model I and Zone C of
Model II, and for magnitudes 7.0 to 8.1 in the Yakataga Zone.
The parameters a and b in the Gutenberg-Richter relations was estimated in each of these
regions from the historical seismicity data.
In addition a characteristic earthquake of magnitude 9.2 with a recurrence
time of 750 years was assumed for the Eastern Zone of Model I and for Zone
C of Model II.
Smaller Earthquakes
Hazard for earthquakes in the magnitude range 5.0
to 7.0 was calculated from a Gutenberg-Richter recurrence relation determined
for each of the megathrust source zones.
a and b- values were determined
for each of the source zones of the megathrust. These seismicity parameters
were determined from maximum-likelihood fits of log N versus magnitude for
shallow events greater than magnitude 4.5, the minimum magnitude of completeness
in the region since 1964. Hazard was
calculated using the a-value and
b-value for each zone. No smoothing
was applied to the edges of these source zones. Details may be found in the companion report
by Mueller et al. (1998).
Active
Crustal Faults
Although considerable information is available about
a few active crustal faults in Alaska, there are certainly many more faults
with unknown slip rates. Faults included
explicitly in the map are shown in Figure 6. (Note that the seismic hazard associated with faults not explicitly
included in the map is captured to a large degree by the smoothed seismicity
model described below.) To be included
in this map a fault must have an estimated slip rate. As in our treatment of western U.S. faults
in the national maps (Frankel et al.,
1996), we divide the faults into two types: A (characteristic) and B (hybrid).
The A-type faults are faults with "known" segmentation. We use a
characteristic rupture model for the A-type faults in which rupture occurs
only as the largest earthquake estimated for each fault segment. B-type faults
have "unknown" segmentation, so we use two equally-weighted recurrence
models. For these hybrid faults, we calculated hazard using 50% weight
for the characteristic earthquake model and 50% weight for a truncated Gutenberg-Richter
frequency-magnitude relation. We used a minimum magnitude of M6.5
and a maximum magnitude of Mchar for the Gutenberg-Richter
hazard calculation. In general, the Gutenberg-Richter recurrence model yields
higher hazard for a given fault than the characteristic model, because of
the more frequent occurrence of moderate-sized earthquakes in the Gutenberg-Richter
model. Use of this model is intended here to account for the possibility that
the crustal faults will rupture in segments smaller than their entire length.
Recurrence times (characteristic) and a- values (G-R) for the each fault were
determined from their slip rates, not from the seismicity surrounding the
fault. Characteristic magnitudes were
determined from fault lengths using the relations of Wells and Coppersmith
(1994). See Frankel et al. (1996)
for the formulas deriving the a-
values and characteristic earthquake rates from the geologic slip rate and
area of fault. The fault widths of crustal faults were estimated by assuming
a maximum faulting depth of 15 km. These crustal faults were taken to be vertical,
except for the Castle Mountain fault (dips 75° to North) and the Transition
fault (dips 10° to North).
Table 1 identifies the faults (and fault segments) included in the map, and
the assumptions made about them. The slip rates are from Nishenko and Jacob
(1990) and Plafker et al. (1993) except as noted below. The recurrence times
in the table are for the characteristic earthquakes. Again, the hybrid faults
also use a Gutenberg-Richter recurrence model which will produce more frequent
recurrence for earthquakes between M6.5 and Mchar.
Estimates of the slip rate for the Totschunda fault
ranged from 8 to 15 mm/yr (Plafker et al., 1993). We adopted a "mean" value of 11.5 mm/yr.
We note that we treated the Transition fault as an
A-type fault, even though its segmentation is unknown. In initial hazard calculations using a hybrid
approach for this fault, we found that the Transition fault produced the highest
hazard in the map. The slip rate of the Transition fault is highly uncertain
(J. Lahr, pers. comm.), and we were concerned that one of the more poorly
known faults had a higher hazard than the megathrust zone. Consequently, we used only the characteristic
rupture model for the Transition fault, which produces a lower hazard than
the hybrid approach. Also, we used the Youngs et al. (1997) subduction zone
interface attenuation relation when estimating ground motions for the Transition
fault (see attenuation section below). This produces somewhat lower ground
motions than do attenuation relations for crustal earthquakes.
Of particular interest is the Castle Mountain fault, passing about 40 km from Anchorage. The fault has been considered in two segments the western, or Susitna segment, and the eastern or Talkeetna, segment (Detterman et al., 1994). Along the Talkeetna segment there is no evidence for surficial displacement younger than Pleistocene (Detterman et al., 1976), but Lahr et al. (1986) describe an earthquake of Ms 5.2 which indicated slip at a depth of 13 to 20 km along the segment. In contrast, along the Susitna segment, no significant earthquakes have been instrumentally located, but geologic studies indicate Holocene surface displacement (Detterman et al., 1974, 1976; Bruhn, 1979). These studies, however, have served only to put wide limits on the slip rate. We chose a value of 0.5 mm/yr and a maximum magnitude of 7.5 leading to a recurrence time of 1300 years (R. Updike, oral communication, 1997).
Determination
of Seismicity Parameters from
Spatially-smoothed
Seismicity
For the smoothed seismicity calculation, the shallow
events (focal depth <50 km) in the areal source zones of the megathrust
were first removed from the catalog. Next the catalog was divided into shallow, deep (focal depth 50-80
km) and deeper (focal depth 80 to 120 km) events. b-values
were determined separately for the shallow, deep and deeper events using the
maximum likelihood method (Weichert, 1980) for events with magnitudes greater
than 4.5. We found b-values of 0.87 for the shallow seismicity, 1.2 for the
deep seismicity (50-80 km depth), and 1.15 for the deeper seismicity (80-120
km depth).
Using the approach of Frankel (1995), a-value grids were calculated using the
maximum-likelihood formula from Weichert (1980). These a-value grids were
then smoothed with Gaussian smoothing functions (correlation distance of 75
km) and the hazard was calculated by summing the frequencies of exceedance
for all of the grid cells. This was done separately for the shallow and two
deep cases. These a-value grids
were then used as the basis to calculate the hazard arising from earthquakes
in the magnitude range 5.0 to 7.0 (5.0 to 7.3 for the shallow case).
Source
Finiteness in Hazard Calculation
In all the calculations of hazard, the treatment
of source finiteness varied with magnitude range. For events between M5.0
and M6.5, we assumed point sources. For events from M6.5 to M7.0 (M6.5 to M7.3 in the case of shallow earthquakes)
we used finite faults of arbitrary strike. For events greater than magnitude
7.0 in the megathrust zone, we used floating rupture zones offset incrementally
along the megathrust.
For the crustal faults, when using the Gutenberg-Richter
recurrence model, we floated the rupture zones along the fault. The floating
rupture zones along the crustal faults and the megathrust cause a tapering
of the hazard at the ends of the faults.
Attenuation
Relationships
The reference site condition is the NEHRP B/C boundary,
which corresponds to an average shear-wave velocity of 760 m/sec in the top
30m. This is the same site condition used in the 1996 national maps. This
site condition represents a typical western U.S. "firm-rock" site.Table 2 shows the ground motion relationships used in the calculations. These
are the same relations used in producing the 1996 hazard maps for the western
U.S. For crustal faults we used different ground motion values for thrust
faults and for strike-slip faults, using the values specified in each attenuation
study. For the deep earthquakes, we assumed a focal depth of 60 km for
earthquakes in the depth interval 50 to 80 km, and a focal depth of 90 km
for earthquakes in the depth interval 80 to 120 km.
Discussion
of Maps
Probabilistic seismic hazard maps for Alaska calculated
as described above are shown in Figures 7-10, for peak ground acceleration,
5.0, 3.3, and 1 Hz spectral acceleration, and for 10% and 2% probability of
exceedance in 50 years.
Hazard is highest in the coastal regions adjacent
to the megathrust and the Transition Fault and in regions adjacent to the
Denali and Fairweather-Queen Charlotte fault systems at all periods and probability
levels. In the interior of Alaska,
away from the Denali fault, hazard is dominated by the spatially-smoothed
seismicity. The region of lowest hazard
in Alaska is along the northern coast adjacent to the Arctic Ocean.
The hazard associated with the Castle Mountain fault is overwhelmed
by the megathrust at most periods and probability levels, but can be seen
on the map of peak ground acceleration with 2% probability of exceedance in
50 years (Figure 7b.) In general hazard in the higher hazard regions
of Alaska is comparable to areas of higher hazard in California (Frankel et
al., 1996).
Disaggregations by magnitude and distance to the
source of the hazard are shown in Figures 11, 12 and 13 for Anchorage, Fairbanks
and Juneau for peak ground acceleration and 5 Hz and 1 Hz spectral acceleration
at the 2%-in-50 year probability level. The disaggregation plots include the summary joint distribution
statistics, the mean magnitude and distance ("mbar" and "dbar"),
and the modal magnitude and distance ( "mmode" and "dmode").
The disaggregations for Anchorage indicate the role
of great earthquakes (magnitude 9 at a distance of about 50 km) relative to
the other sources (Figure 11a, b,
c).
The relative contribution of the great earthquakes increases with increasing
period (decreasing frequency) until the great earthquakes dominate at a frequency
of 1 Hz. On these plots hazard from
the Castle Mountain fault at a distance of about 40 km is combined with the
hazard from earthquakes in the magnitude range up to 7.5 on the megathrust.
The other significant contributions to the hazard
arise from the shallow smoothed seismicity (shown at a distance of about 20
km) and the two deeper zones of smoothed seismicity ( show at distances of
60 and 90 km).
The disaggregations for Fairbanks (Figure 12a,
b, c)
show that the hazard is dominated by local earthquakes. The influences of the Denali fault at distance of about 135 km and
the megathrust at a distance of about 350 km are only apparent on the disaggregation
for 1 Hz.
The disaggregations for Juneau (Figure 13a,
b, c)
show the relative contributions of the local shallow seismicity and large
earthquakes on the South Denali (Chatham Strait) fault at a distance of about
80 km. The relative contribution of
the large earthquakes on the South Denali increase with increasing period
(decreasing frequency).
Although somewhat more detailed, the current maps
for peak ground acceleration are generally similar to the maps prepared by
Thenhaus et al. (1985) for the corresponding probability levels, although
significant differences do exist. (Thenhaus
et al. did not estimate the hazard in the area of the Aleutian Islands.)
In general the values on the current map for the 10%-in-50-year probability
level are somewhat lower than those on the Thenhaus et al. map in south central
and southeast Alaska, but somewhat higher in the offshore regions above and
adjacent to the megathrust. The values
on the current 2%-in-50-year map are similar to those on the Thenhaus et al.
map in south central and southeast Alaska, but somewhat higher in the offshore
regions above and adjacent to the megathrust.
For example, for the Anchorage area at the 10%-in-50-year probability
level, the current map indicates a peak ground acceleration of 37% g, in contrast
to about 45% for the Thenhaus et al. map.
At the 2% in 50 year probability level, the current map indicates about
65% g for the Anchorage area, as contrasted with ~67% g on the Thenhaus et
al. map. Table 3 compares estimates of the estimated peak ground accelerations from the
current maps and the Thenhaus et al. maps at nine locations in Alaska.
Conclusions
and Issues Requiring Future Work
Alaska has some of the areas of highest seismic hazard
in the United States. In contrast
to California where most of the regions of highest hazard occur in relatively
narrow zones and are often associated with nearly vertical faults, most of
the hazardous regions in Alaska occur in association with relatively shallow
dipping faults leading to much larger affected areas. The principal sources of seismic hazard in
Alaska are the Alaska-Aleutian megathrust and the Transition fault (both relatively
shallow dipping and affecting very large areas), and the Fairweather, Queen Charlotte and Denali faults
(near vertical faults leading to relatively narrow zones of high hazard.)
Obviously, there are many aspects of the methodology
and input information that can be debated. In constructing these maps we have attempted
to include those elements of geologic and geophysical information and interpretation
for which a community consensus exists. We
have also attempted to indicate explicitly those areas where we have been
forced to rely on judgment or assumption.
Future seismic hazard maps would benefit from additional
information in several areas. Additional
understanding of the characteristics of faulting in subduction zones, especially
the 1964 zone would be very helpful. Tectonic understanding of the region between
the 1964 zone and the Fairweather fault, including the Yakataga gap and the
Transition fault would be extremely valuable.
More measurements of slip rates on the crustal faults are needed. Finally, new insights into the segmentation
of the megathrust and the expected magnitude distribution of earthquakes would
have a large impact.
Specific methodological issues that warrant further
consideration include: 1) Segmentation
boundaries on faults characterized by large earthquakes cause saddles in hazard
with lower hazard near boundary, because of the tapering of hazard caused
by floating rupture zones. We dealt with this saddle in the hazard in an ad
hoc manner by overlapping the zones. This issue requires additional consideration over the long run.
2) We used historical seismicity rather than convergence rates to establish
rates of large (M>=7) earthquakes in the megathrust source zones.
It would be desirable to rationalize the seismicity rates with the
convergence rate through some quantitative mechanism. 3) Future segmentation models
for the megathrust could benefit greatly from additional tectonic insight.
4) Use of time-dependent probabilities.
Acknowledgments
We thank the participants of the September 1996 workshop
on seismic-hazard mapping in Alaska for their valuable suggestions and guidance.
Many of their comments were applied to make the draft maps. We particularly
thank Max Wyss and Stefan Wiemer of the Geophysical Institute and John Lahr
of the USGS for their help with the treatment of earthquake catalogs (see
accompanying documentation). George
Plafker, Randy Updike, and Don Wells provided useful discussions.
C.B. Crouse, K.H. Jacob and E.V. Leyendecker provided helpful reviews.
Ken Rukstales prepared the full color 1:7,500,000 scale maps for publication.
References
Boore, D., W. Joyner, and T. Fumal
(1997). Equations for estimating horizontal response spectra and peak acceleration
from Western North American earthquakes: a summary of recent work, Seism.
Res. Letts., v. 68, no. 1, pp. 128-153.
Boyd, T.M., E.R. Engdahl, and W. Spence,
(1995). Seismic cycles along the Aleutian
arc: Analysis of seismicity from 1957
through 1991, J. Geophys. Res., v.
100, pp. 621-644.
Bruhn, R.L. (1979). Holocene displacements
measured by trenching the Castle Mountain fault near Houston, Alaska, in Short
Notes on Alaska Geology, Alaska Division of Geological and Geophysical Surveys,
Geologic Report 61, Anchorage, Alaska, 4 p.
Campbell, K. and Y. Bozorgnia (1994).
Near-source attenuation of peak horizontal acceleration form worldwide accelerograms
recorded from 1957 to 1993, in Proceedings of Fifth U.S. National Conference
on Earthquake Engineering, Earthquake Engineering Research Institute, Oakland,
California, III, pp. 283-292.
Cormier, V. (1975). Tectonics near the junction of the Aleutian
and Kuril-Kamchatka arcs and a mechanism for the middle Tertiary magmatism
in the Kamchatka basin, Geol. Soc. Amer. Bull., v. 86., pp. 443-453.
Cornell, A. (1968). Engineering seismic risk analysis, Bull. Seism.
Soc. Am., v. 58, pp. 1583-1606.
Davies, J.N., L. R. Sykes, L. House,
and K. Jacob (1981). Shumagin seismic
gap, Alaskan peninsula: History of great earthquakes, tectonic setting and
evidence for high seismic potential, J.
Geophys. Res., v. 86, pp. 3821-3855.
Detterman, R.L., G. Plafker, T. Hudson,
R.G. Tysdal, and N. Pavoni (1974). Surface geology and Holocene breaks along
the Susitna segment of the Castle Mountain fault, Alaska, U.S. Geol. Surv.
MF Map 618, 1 sheet.
Detterman, R.L., G. Plafker, G.T.
Russell, and T. Hudson (1976). Features along part of the Talkeetna segment
of the Castle Mountain-Caribou fault system, Alaska, U.S. Geol. Surv. MF Map
638, 1 sheet.
Frankel, A. (1995). Mapping seismic
hazard in the Central and Eastern United States, Seism. Res. Letts, v. 66,
no. 4, pp. 8-21.
Frankel, A., C. Mueller, T. Barnhard,
D. Perkins, E. Leyendecker, N. Dickman, S. Hanson, and M. Hopper (1996). National
seismic-hazard maps: Documentation June 1996, U.S. Geological Survey Open-File
Report 96-532, 110 pages.
Lahr, J.C., C.D. Stephens, H.S. Hagewara,
and J. Boatwright (1980). Alaska seismic gap only partially filled by 28 February
1979 earthquake, Science, v207, pp. 1351-1353.
Lahr, J. , R.A. Page, C.D. Stephens,
and K.A. Fogelman (1986).
Sutton, Alaska, earthquake of 1984:
Evidence for activity on the Talkeetna segment of the Castle Mountain
fault system, Bull. Seism. Soc. Am., v. 76, pp. 967-984.
Mueller, C., A. Frankel, J. Lahr,
and M. Wyss 1998). Preparation of earthquake catalogs for the national seismic
hazard maps: Alaska, U.S. Geological Survey Open-File Report, in preparation.
Nishenko, S.P., and K.H. Jacob (1990).
Seismic potential of the Queen Charlotte-Alaska-Aleutian-seismic zone, J.
Geophys. Res., v. 95, pp. 2511-2532.
Pacheco, J., L. Sykes, and C. Scholz
(1993). Nature of seismic coupling along simple plate boundaries of the subduction
type, J. Geophys. Res., v. 98, pp. 14,133-14,159.
Page, R.A., N.N. Biswas, J.C. Lahr,
and H. Pulpan (1991). Seismicity of continental Alaska, in Slemmons, D.B.,
Engdahl, E.R., Zoback, M.R., and Blackwell, D.D., eds., Neotectonics of North
America: Boulder, Colorado, Geol. Soc. Am., Decade Map Volume 1, pp. 47-68.
Plafker, G., L. Gilpin, and J. Lahr
(1993). Neotectonic map of Alaska, in Geology of Alaska, vol G-1 of Geology
of North America, Geological Society of America, Boulder, CO.
Plafker, G. and M. Rubin, (1994).
Paleoseismic evidence for "yo-yo" tectonics above the eastern
Aleutian subduction zone: coseismic uplift alternating with even larger
interseismic submergence, in Proceedings of the Workshop on Paleoseismology,
C. S. Prentice, D.P. Schwartz and R.S. Yeats, Convenors, USGS Open-file report
94-568, p.155-157.
Sadigh, K., C. Chang, J. Egan, F.
Makdisi, and R. Youngs (1997). Attenuation relationships for shallow crustal
earthquakes based on California strong motion data, Seism. Res. Letts., v.
68, no. 1, pp. 180-189.
Sykes, L.R. (1971). Aftershock zones of great earthquakes, seismicity
gaps, and earthquake prediction for Alaska and the Aleutians, J. Geophys. Res., v. 76, pp. 8021-8041.
Sykes, L.R., J.B. Kisslinger, L. House,
J.N. Davies, and K.H. Jacob (1981). Rupture zones and repeat times of great
earthquakes along the Alaska-Aleutian arc, in Simpson D.W. , and R.G. Richards,
eds., Earthquake prediction: American
Geophysical Union, pp. 73-80.
Taber, J.J., S. Billington, and E.R.
Engdahl, (1991). Seismicity of the
Aleutian arc, in Slemmons, D.B., Engdahl, E.R., Zoback, M.R., and Blackwell,
D.D., eds., Neotectonics of North America: Boulder, Colorado, Geol. Soc. Am.,
Decade Map Volume 1, pp. 29-46.
Thenhaus, P.C., J.I. Ziony, W.H. Diment,
M.G. Hopper, D.M. Perkins, S.L. Hanson, and S.T. Algermissen (1985).
Probabilistic estimates of maximum seismic horizontal ground acceleration
on rock in Alaska and the adjacent continental shelf, Earthquake Spectra,
v. 1, pp. 285-306.
Youngs, R., S. Chiou, W. Silva, and
J. Humphrey (1997). Strong ground motion attenuation relationships for subduction
zone earthquakes, Seism. Res. Letts., v. 68, no. 1, pp. 58-73.
Weichert, D. (1980). Estimation of
earthquake recurrence parameters for unequal observation periods for different
magnitudes, Bull. Seism. Soc. Am., v. 70, pp. 1337-1356.
Wesson, R.L., A.D. Frankel, C.S. Mueller,
S.C. Harmsen (1999). Seismic-Hazard
Maps for Alaska and the Aleutian Islands.
U.S. Geological Survey Miscellaneous Investigations Series I-2679,
2 sheets.
Figure Captions
Figure
1. Instrumental seismicity of
Alaska and the Aleutian Islands from the consolidated catalogs.
Earthquakes shown have magnitudes, Mw>=5.5, and dates ranging
from 1880 to 1996. (Mueller et al., 1998).
Figure
2. Rupture areas of large earthquakes in the Alaska Aleutian region
during this century (Plafker et al., 1993).
Note that virtually the entire boundary between the Pacific and North
American Plates has ruptured during this period with the exceptions of 1)
the western most Aleutians (~168°E), 2) the Shumagin gap (~160°W),
and 3) the Yakataga gap (~142°W). See
text for discussion.
Figure
3. Hazard Model for Alaska.
Figure
4. Segmentation Model I for the Alaska-Aleutian Megathrust
Figure 5.
Segmentation Model II for the Alaska-Aleutian
Megathrust
Figure 6.
Active crustal faults identified in
Alaska (Plafker et al.,1993).
Figure
7a. Peak ground acceleration (%g)
with 10% probability of exceedance in 50 years.
Figure
7b. Peak ground acceleration (%g)
with 2% probability of exceedance in 50 years.
Figure
8a. 5 Hz acceleration (%g) with
10% probability of exceedance in 50 years.
Figure
8b. 5 Hz acceleration (%g) with
2% probability of exceedance in 50 years.
Figure
9a. 3.3 Hz acceleration (%g) with
10% probability of exceedance in 50 years.
Figure
9b. 3.3 Hz acceleration (%g) with
2% probability of exceedance in 50 years.
Figure
10a. 1 Hz acceleration (%g) with
10% probability of exceedance in 50 years.
Figure
10b. 1 Hz acceleration (%g) with
2% probability of exceedance in 50 years.
Figure
11a. Disaggregation of hazard
at Anchorage for peak ground acceleration at 2%-in-50 year probability level.
Figure
11b. Disaggregation of hazard
at Anchorage for 5 Hz ground acceleration at 2%-in-50 year probability level.
Figure
11c. Disaggregation of hazard
at Anchorage for 1 Hz spectral acceleration
at 2%-in-50 year probability level.
Figure
12a. Disaggregation of hazard
at Fairbanks for peak ground acceleration at 2%-in-50 year probability level.
Figure
12b. Disaggregation of hazard
at Fairbanks for 5 Hz ground acceleration at 2%-in-50 year probability level.
Figure
12c. Disaggregation of hazard
at Fairbanks for 1 Hz spectral acceleration
at 2%-in-50 year probability level.
Figure
13a. Disaggregation of hazard
at Juneau for peak ground acceleration at 2%-in-50 year probability level.
Figure
13b. Disaggregation of hazard
at Juneau for 5 Hz ground acceleration at 2%-in-50 year probability level.
Figure
13c. Disaggregation of hazard
at Juneau for 1 Hz spectral acceleration
at 2%-in-50 year probability level.
Table Captions
Table 1.
Characteristics of active faults assumed
for hazard analysis.
Table 2.
Attenuation models assumed for various
seismic sources assumed for hazard analysis.
Table 3. Comparison of peak ground acceleration
for selected locations in Alaska between the current maps and those prepared
by Thenhaus et al. (1985).