ML20151H253
| ML20151H253 | |
| Person / Time | |
|---|---|
| Site: | Trojan File:Portland General Electric icon.png |
| Issue date: | 04/12/1988 |
| From: | Hartzell S, Heaton T AMERICAN ASSOCIATION FOR THE ADVANCEMENT OF SCIENCE |
| To: | |
| Shared Package | |
| ML20151H012 | List:
|
| References | |
| NUDOCS 8808010201 | |
| Download: ML20151H253 (8) | |
Text
vm u - - - -
. - - . . - . - . . ~ . . . . ,
e- @
(.
10 april 1987, Vohme 236, pp.162-168 SCIENCE 4
/
c Earthquake Hazards on the Cascadia Subduction Zone Tuoms H. HEATON AND SMHEN H. HARTZEU.
Cnpyright c 1987 t,y the American Anociation for the Advancement of Science 8800010001 000615 3 DR ADOCK 050 4 f
( Earthquake Hazards on the Cascadia Subduction Zone THOMAS H. HEATON AND STEPHEN H. HARTZELL 700,000-year old Brunhes Maruyama magned rewrsal. Nishimura large subduction earthquakes on the Cascadia subduc. er 4. (2) calculated a converFence rate of 3.5 to 4.5 cm' yen vross tion zone pose a pntential seismic hazard. Ver7 young the Cascadia suWuction zone from sea-floor magnetic lineation oceanic lithosphere (10 million years old) is bemg sub- data. He suitabibty of this kinematic model to present day plate ducted beneath North America at a tre of approximate!y monons is supported by work of Hyndman and Weichert (3) who 4 centimeters per year. Le Cascad2a subduction zone showed that historic scismicity is compatibk with shp rates espected shares many characteristics with subduction zones in from magretic hneation data or' all the plate boundaries of the southern Chile, southwestern Japan, and Colombia, Pacific Northwest except on the Cucadia subduction ume. It seems where comparably young oceanic lithosphere is also sub- afEcult to corutruct a nniel ofplate mooons that shps 5 cm' year on ducting. Very large subduction earthquakes, ranging in plate boundaries both north and south of the Cascadia subduction energy magtutude (M.) between 8 and 9.5, have occurred zone, but with no convergence on the subduction zone.
along these other subduction zones. If the Cascadia subduction zone is also storing :lastic energy, a sequence cf several great carthquakes (M. 8) or a giant carthquake (M 9) would be necessary to fill this 1200 kilometer ga Subduction of Youn8 LithosEhere he nature of ctrong ground motions recnrded durm, p. g ne oceanic Uthosphere that is subducting beneath the Pacific subduction earthquakes ofM. less than 8.2 is discussed. Northwest is very young, about 10 milixm yeus old (4). Seseral Strong ground motions from even larger carthquakes chuacteristics of the Cascadia subduction zone distinguish it from U, np to 9.5) are estimated by simple simulations. If most other uorldwide subduction zones, and most of these can be
.arge subduction earthquakes occur in the Pacific North- att_ibuted to the youthfulness of the subducted oceanic bthosphere.
west, relatively strong shaking can be expected over a More specifically, there is no significant bathpnetric trench or large large region. such carth< J. ts may also be accompanied gravity anomaly for the Casca6a subduction zone. He sea floor by large local tsunamis, adjacent to the continental margm is only about 3 km deep, and the average heat flow is relatively high (5); both of these features are directly attributsble to the ymth of the oceanic bthosphere (6).
Furthermore, although there is a 6stinct Benioff.Wadan seisnucity rstrrE CoMPEt11NG EY1MNCE ntAT Tilr GoRDA,itJAN zone i.I the Paci6c Northw est (7),it stops at a much shauower depth D de Fuca, and Empk>rer plates ve actively subductmg ak>ng than in nut cxhe subduction zones, extendmg to a depth of less the 1200Lkm kmg Cuca6a subducuon rone in the PaaSc than 80 km.
Northwest, there have been no luge hatoric shallow subduction Uyeda and Kanarrori (8) suggested that the scismic couphng of earthquakes of the rge cigrienced at most cxher subductmg plate subducting plate tuundaries (the fracnon of plate shp that cecun bounduies. Is the Cuc4&a subduction proceu benign, with the during cuthquakes) is related to the physical characteristics of the 6fftrential plate motion occurnng through uei mic creep, or is the plate tvunduy. Ruff and Kanamori (9) derrorutrated that wcakly tone stonng clutic strain energy to te relemed in future great coupled zones tend to have slow subduction of very old oceang subduction euthquakes? If the Casca&a subduction une b noring crust, wheren strongly coupled renes tend to have fut subductkm strain energy, how luge might the cuthquakes be, how often might of young crust. ney suffested that old hthosphere is dense and they occur, and what might the ground motxms be? nese are all subducts spontaneously with oceanward retreat of the trench and 6f5 cult, but vital, questxms wtue answers dramatically affect the subsequent ogning of back arc bains (Mariana type, weakJy sstirnatxm of scumx nsk in the Pacifx Northwest. coupled). Young, buoyant bt}msphere tends to subduct (mly w hen it is overridden by continental bthosphere, as is the cue akeg much of the westem cout of North and South Arnenca (Oilean type, Active Subduction in the Pacific Northwest ""*8'Y ' "P 3'd)'
he pometry of the major plate boundaries and seisnsity in the Pmfx Northwest is shown in Fig.1. A half spreadmg rate of 3 .
cm'yeu wu inferred for the luan de Fuca ridge by Delaney rt 4 (1) Co arison of Cascadia with Other ho reported that 43 km of new oceanic cnut hu fornwd since the Su uction Zones Heaton and Kananwm (10) pointed out that the Casca&a subducton ume hu phyucal charaaensucs wry afferent from b ksEn52 D E [A .
Cauma insnrute rd %.p, rasaima. CA 91125 ima 1I those of the tra&tiona! *pcismic" subduction une (weakly coupled, Mananas type). Furthernwre, they reported that th? Cuca&a
'e SClfNCI,voL a%
~
. c ._
- .r,
( 3- - --.--w 54,f 2? . & 4 "
. ' 2,( '".'
.n. .. -.s .
.. o .
. .g,. .-
- h. .
e'v-<
u h. . . -
. L
. \,, ...m . ,
\
.4 ' *
- W , _ .__
'~f
~r s' pe s ....mi- .:
- 8
+ - -
\..
$ +. ,,:: r %.. ,
. T'.'
.lQ"\"
V.
\i ..
=; . . .-
\. -
e .. y I i t'x. % ... . . s. :. s
.-- . = -
. 4_._A~
- . ./ .
l
. .. s .
I. ,
o as s
.uc s *
',N s . .
,.=
o,egon
- 1 ..
. .- A I
- y. .y.
Q.
.=:. .
l . 1 *
~ ff . , lw n-
. ~* . . .*.' .:, :. o
. . . w 1 .
- e
\, \.
I.* ..
e.t b .
4 t=, ,
- \.
\
t
. \ . -),:
- 1 .
~ -. ,,, ~3. ._. . . ,
l .. -
e.< , o l\
- . ..: r 9
\..
-..,:* , . p.W.
.n '
. .s . 4 . .f.
i . .
, Q b - , -. '
f .e . ..
N< m - a y ..* ,r 1 .
J. .t::':.:.. _. . .t . .- -
.' ... i ::,. ;.. .- . . .u. J .s ..
. e.-l
- s. w ~r r
___,,p,;;,. .s,r-,r.R.;,s. ,
8 2 I4 "*F" M -
R. tat .e fastt went
. .. .. - n. . ~,
- * * * - " " ~ - %.. c.,,,e
. 4 0 - 4 9 u.r.tude f
1
\
l f 1 J
' O 3M km Fig.1. $ctirmary and plate teswtio of the Pacdx Northwest.
q to APML t98? ART 1CLil 16) 4
________,.-,_,~.-.-.....,.------_,--_-,,,.__._,,_,.-,___--,m.,-.-- -
.-....-,-.,_~.~_,,-y , , - . - , , , - .
-4y -4& -79* -70'
-36* -35'
_ , y
/ **_
.g i
g . -iry
\,, , , ,0 et Q f , .
. . -130'
. . - . . -175*
- k. 8
\ - pap,. .
e,,A.m*l',%:.'._ ,
4- %
. ,, ra xA., s g4 .
-e -
,_. e ,,q ' g ,. -e _ .
. en
\, ,. p j p gI ,. :. -- . -
%= j ,N.y,_. j ,a ,
kw g M'.\ "
k1 _
f I a . .
' l )g , .
~
) / /' - \,
(
\ . , d,1 g
- n. ,. -
,a
\\
r> -r/ pi- r -
Nif j/ 'A /;i .
\\ '\ '\ * ~ . , r
\ y k hb. of q ,
's' ase \ \\ S . aSe 8sy
-gp1.T, \, \. , r 'f '\
- Qsc~
\ \ - Q j) ^ - ;
I ?
.j.h.$- I
? /II' l
-. -- ,s, .
4 se . \\ % - . 1 _
\\
q st sa ,)- *'!*.?L*9.f.f , ,.,
_ m , . , ,
e I L \ . -130' -125' -120'
.,,. , , , ,\, '
,,y J , .
-4 5* -40'
-75' -70*
I m 35W ,
1FE - Y '
~
_ c " M. D't f
i m-. t j
~
- 'I
.,, Fig. 2. Comparuon of subductxm in southem Chile ( A), racifx Northsest
'.. 7 ed on approumately the N.. A (B),
sameand southwes Active : aemg laban (C). A11 trups are
_/ sc
.# p . h" . sC ~ *
- f. ]
r ,
scale.
kneamons (bght centers (hean hnes), sca&x magneuc hnes), Quateman vokanoes (sobd enangles), and l fracture zones (F.Z.) are shown. Spnbob on the magnene kneauens refer to
$$d: g e/ sa,/
/ i 7'" r iAd)\
I the geomagnetw eme scale; hneation J is about I mdbon years,5 about 10 mdbon years,6 about 20 'nuon years, and 10 about 30 mdbon years. The appr umare rupture areas of large Japanese earthquakes and contouri 135,E -
'M _
935.E e
N
, *[g
~ ""
y (rneters) of sertxal deformanon assocated etch the 22 May 1960 Oulcan earthquake are also sho.m (n).
n $"s j-~ / "p/ N '
k ,,, 9 ,
l l
- - ;m A - l 5 < .i ,- .*
, . ,, 1 140*E i ! i ' - 1 m 3sw subduction zone ha many similarities to other strongly coupled the same scale in Fig. 2, wtuch shows the similarity of the overall subduction scoes (Osilean type). Heaton and Hart:cu (11) con- damensions of these plate bos.nciaries.
ciuded that of a!! woektwide subduction zones, thermG subduc. Kanamori and Astia (12) have suggested that a the age of the tion zone seems most similar to those in southem Otile, southwest. subducted plate approaches 0 years, we may expect increasing em Japan, and Colombia. Very young occarde lithosphere is M anewmc slip to result from high temperatures at the subducting 4
f ducting in each of these locatic.u, and very large, shallow, thrust boundary. Perha the Cascadia subducuan zone is ao hot that slip
\ earthquakes have occurred at each of these wher zones. Maps along this is occurring u ancistnic creep. Unfortunately.
Comparing the geometries of the subduction zones tri che Pacific we do ti. t uw the age at wtuch this mechanism may become Northwest, southem Odle, and southwestem Japan are shown on importan: )lowcwr, at other locaticnt where the pungest occarde 164 SCl!NCE, YOL. 2)6
~ --.=.. ..
lithosphere is subducting (less than 5 million yean in southem Howner, such feamtes are relatively rare since they are presene Gile) (Fig. 2), major shp has occurred during great carthquakes only when there are high, long term upbft rates, luge coseismi (11). uphhs, and moderate to low coastal erosion rates. From rniews of
-( One of the most remukable features of the Cascadia subduction the spane Uterature on Pleistocene marine terraces, Adarns (1 zone is the striking paucity of historic shallow coutal seismicity, reprted relatively slow emergence and possib!c submergence for Although this may indicate aseismic slip along the plate bounduy, rnost of the coastline of Wuhington and northem OreFon. He the sections of the San Andreu fault that are currently creeping have further reported moderate upbft rates ofless than 1.5 mm' ye a
a relatively high level of small earthquake seismicity, whereas the southem Oregon. Coastal crosion rates are high, and fe secoons of the fault that experienced peat earthquakes in 1857 and uplifted Holocene strand Enes have been identined along the cou 1906 are currently devoid of any meuurable actbity. One explana. of Washington and Oregon. The highest geologic uphft rate (3 tion for this behavior is that stress increues smoothly and uniformly cm' year) documented along the Cascadia subduction zone occurs ,
on fault zones that we coupled over luge ares, wheren numerous near Cape hiendoesno, Cahfornia, in the region of the Gorda-stress concentrations occur on faults haing large areas that underFo Paci6c-North America triple junction (20). A Bight of nine emer.
acistnic sbp. Heaton and Hartzell (11) noted that signi6 cant gent terraces and beach tidges ha formed during the put 5000 p riods oflow seismicity have been observed at subduction zones yean at Cape Mendocino (20), but it is uncertain whether these similar to the Cucadia subduction zone, whereu subduction zones terraces have any beanng on the problem of luge subduction of tM Mariana type (mainly aseismic slip) show a low, but steady, earthquakes in the Paci6e Northwest.
rate of seismicity. Heaton and Hartzell (11, pp. 697-698) stated The lack of raised Holocene terrace in Oregon and Wahington that "although [the compuison study) does not prove that peat may be due to cosessmic coastal subsidence, a commonly obsen e carthquakes will occur on the Cucadia subduction zone, it dvea typtinfuct of great subduction earthquakes. Most of the contal suggest that it is inappropriate to usume that great cuthquakes will area adjacent to the 22 May 1960 Chilean earthquakes sub not occur based on obse.vations of bathpnetry, hthospheric age, I to 2 m, w hereu uphft occurred only at the outer islands and at the trench se&ments, heat Bow, convergence rate, physiopaphy, overall nonhem end of the rupture zone (Fig. 2). In a reconnaiss size of the subducted plate, Quatemuy vokanism, or the rate of of Hokicene relative sea levels on the Weshington cout, Atwater backpound seismicity."
(21) found nidence that a tidal manh subsided suddenly neu Neah Bay (northwestemmost tip of Wuhington) approximately 1100 yean ago. He also found similar niderxe for multiple jerks of ,
Cascadia Subduction Zone-Locked or mbsidence in the Walapa Bay region 200 km to the south. Atwater ;
Unlocked? (21) P' F"5'd th '"h Ik f 'ubsidence may muk a peat prehistoric cuthquake on the Cuca&a subduction zone.
Although comparison of the Cucaia subduction zone with other Holume rar6diter. Adams (f,22) ha suggested that extenshe subduction zones leads us to bebne that there may be a potential for Holocene turbi6tes studied by Gnggs and Kulm (23) large subduction euthquakes, it would be far more satisfying to been triggered by luge euthquakes along the continental sh hac both & rect nidence that clastic strain is accuinulating and Stnce the deposition ofMuama nh 6600 yean ago, therc hu e nidence for prehistonc targe earthquakes. Since the question of approxima+ely 16 major turbi&tes in the Cuca6a Channel, w hich is luge subduccon euthquakes hu been asked onh recently, the consistent wth an average euthquake repeat time of 410 search for & rect nidence is std1 in a very culy stage.
Adams (22) pointed out thu wo sepuate channels separated by 50 Geodite irram. Savage er 4. (13) and later 1.isowskj and Savage km, and that also feed the main Cucaia Channel, have turbidite (14) 63 cussed Geodohte surveys made in the Seattle region from sequencea compuabk in number to those seen in the main channe 1972 to 1986 and triangulation surveys along the Strait of Juan de 71us seems to be nidence that turbiary currents were simu Fuca from 1892 to 1954. Their analyses in6cate that both regions ly tnggered in separate channels, perhaps by peat sub show a muimum contraction in a direction that is neuly pualJc! to cuthquakes.
the cut northeast plate convergence directons at rates of Huroned recordt. Although fint explored by Europeans in the late 0.03 2 0.01 and 0.2 2 0 07 microstrain per year for the Scutic and 1700s, coastal Wuhington had no permanent Caucasian set Strait of Juan de Fuca regioru, respectively. Unfortumtely, ambigu. untd 1810. There being no known wriner. :ccounts o ity exists in the interpretation of the strains from the Seattje Geodobte network, at leur putly because of a poor signato-noise that can be interpreted as a peat subducuo ratio (15). Repeated inehng surveys along much of the couthne unlikely that any such nents hue occurred for a Hown er, a few legends of Wuhmpton contal In6ans suggest the !
i adpcent to the Juan de Fuca plate show uphft of the cout regions at occurrence of a larFe tsunarni along the nortj a rate of up to 3 mm' year and subsidence of the inner coastal area at coast (24). There are aho kFends ofluFe euthquakes and dastur.
a rate of about I rnm' yen (16), l.isowski and Savage (14) showed .
baixes of the contal waters from contal Indiam in northemmost !
that the combined Getdobte, tnangulation, and lotling data can be Cahfomia (24). Unfortunately, these kgends are tco explaned by a rnodel in wtuch the shallow thrust tone is kded I between the trench atis and the coasta! egion. Concismk extensions consutute proof that larFe subduction cuthqu '
ranging from 25 to 60 rrucrostrain were obsened in the central valky o(southern Chde for the 22 May 1960 Chikan earthqude (17), and cosetimk catensions from the 28 March 1964 Aluhn HFPothetical Subduction Earth9 uakes
- earthquake ranged from about 15 rrucrostrun in the inner contal If the Cucadia subduction zone is kded, what sert of earth-areas to more than 50 microstrain in the outer coastal areu (18). At
,f the current strain rates, it would take entral hundred to a thousand quakes may occur there? Ahhough thu question i 2 5 useument of seisnde huards, at this potnt our ams en are sycula-yean for comparable strairu to accurnulate along the Cuc46: ove. For simpbcity, we usume that earthquakes on the Caxa
- subduccon zone.
subduction zone may resemble carthquakes on the subductxm rones Holamr shchers. Hokrene geomorphic and depositional fea.
, tures often record the cmence of peat subduction earthquakes. that seemJapan, southem Chile, southwestem to beand the trnst sinular Cokimbla. Heaton andto the Cuca to APR!t 198*
AILTICt t s 161
g 'usuwo. to u te, owere so se too w un.co.4 g u.e,mo.7eoez o.uic.seeieo w me..a
- 7, ,
7-- , -
{\ %/ / [ 'N 'N/
( l, y?-
/ 4
>+
j/ !
7
? !
/x N
x
[\ / N /.
\ \ \
- ,- ,r
/n gg ,, 5'#y
- t s , , .
{i /. X i % .
t
/ L hs/'\ !' ?
2 /\ f' f f N x/ ,
[ /x .I \
\
[\
v 3
f Nf / \
, r , , u . . . 3 r , , . . . .<, e , a . . . . .<, e , , , , , ,,,
g
) x/ x/ N
, e i , . . . . . . , r . . . . . . . <, e . , , 4 , , , ,,3 e . , , . . . . .
Po%d (e.c)
Fig. 3. Pseudosckwity respome sptra (5% darnped) of horuontal components of ground moton for sites at 6 stances betw een 50 and 100 krn tue large sub-den earthquales.
Hutzell (11) summuized historie canhquake acthiry for these the South Chile subduction zone (Fig. 2). The 6memions of the subduction zones and chose several historic euthquake sequences 1960 rupture zone are compuable to those of the entire Cascadia that ther considered to be plausible for the Pact 6c Northwest if, in subducnon zone (Fig. 2), and thus wt comidu the 1960 canhquake fact, the subduction zone is locked. In the 6nt scenario, a sequence to represent the largest canhquake feastb!c in the Paci6e Nor hwest.
, of four or 6se euthquakes similu to the 1944 and 1946 energy The convergence rate in southem Chik is atout 9 cm' yen, and large
! magnitude (M,) 8.1 southwestem Japan enthquakes (Fig. 2) cathquakes also occurred in this region in 1575,1737,1837, and 1 would be sutTwient to coser the length of the Cascadia subduction 1960-an average recurrence time of about 128 yean. Howe er, rone. Such canhquakes might be closely spaced in time, as ha been there is strong evidence that aserage dislocations of greater than the case for many sequences in southwestem Japan. Plate comer. 20 m accompanied the 1960 euthquake (17), and thus a recurrence
, cence rates in southwestem Japan are compuable to those in the interval of 128 yens seems inconsistently short with respect to the Paede Northwest. The average euthquake recurrence intenal in long term comergence rate. There is oidence that the 1960 euth-southwestem Japan is about 180 yeus (25). quake may have been sigm6cantly larger than presious historic Another scenano calls for an euthquake similu to the 1906 M. nents in this nea (11). If earthquakes sundu to the 1960 Chilean
, 8.8 Colombian earthquake. An average dislocation of about 5 m earthquake do occur on the Cascadia subduction zone, their average oser a rupture kngth of atuut 500 km has been estimated for this recurrence interval would probably etceed 500 yean.
e ent (26). The secuon of the zone that ruptured in 1906 seems to have ruptured again in a sequence of sma!!ct canhquakes (1942, 1948, and 1979) whose moment sum is only one 6fth that of the Rupture Process for Large Subduction 1906 event (Jo). The comergence rate at Colombia is abc,ut 8 cm year, but the historic records are insut5cient to allow an estima. Earth9uake' r
tion of the reorrence intenal of great Colombian canhquakes. We have argued that the ery young age of the subducted i The 22 May 1960 Chilean earthquake is the largest documented hthosphere in the Paci6c Northwest taay determine the seismic I earthquake of this century (N. 9.5). The rupture covered a length coupling of the plate boundary. Is there a systematic difference with 1 of about 1000 km and ruptured through many 6tierent segments of the age of the subducted plate in the nature of seismic energy release during large subduction earthquakes! HartzcIl and Heaton (27) y ,,, stuaed broad. band teleseisnuc P. waves from 63 of the largest e
1 4
Im ;
8.7 {n g
w .***
..,. shallow canhguakes in the lut 45 yean. The earthquakes nuaed ucurred in 15 subduction rones with a wide range in the ages of i I . , , - , . subducted lithosphere and represent a wide range of convergrnce rates and muenum size of earthquakes. Hutzell and Heaton
)
f "***' derived teleseismic time functiom in the peticd band from 2 to 50 l
} ,,
N ,,, , seconds and charxterized those funct>ons in terms of toughness, l
{u -
,_y es we w cwerall duration nuhiplicity of sources, spectral slopes, and dura-g
- ql;*,, tion of indn4 dual pubes. These rneuures vaned widely from one earthquake to another, although carthquakes withm the same
,, q subduction zone seem to be sarrular. Comparing the time functions i e m am se e ,a sm 30 with sp. rate, and matimum M,. nf the sitMartww) fnnes does not
- '*" yicid obsious global trends. These observations indicate that inher-I ( Fig. 4. Compamnn of preactom of nenge peak ground motoru obeatned A reFresuon ana!Ws of data fnxn the southwestem United States (.% and ent 6trerences are nc>t expected in the nature of energy release from from Japan (.U). Sod utes and 5% dampu,g ue u umed for the reyanic carthqt.ies n' subduction zones that are sinular to the Cascadia nptra subduct)on zone.
j 166 &CIINCI, VOL, a)6
__. c - .- __ _ _ ,m .
~
4 - - - .-
Strong Ground Motions Observed During s,-. ... " ~ G,,, ,7 f
Large Subduction Earthqmkn ca.e, f,..j.../.../.../.../
? * * */ !
7/4 _ f *4 Hyp .<eme,
'( Heaton and liartzeu (28) ascussed the nature of strong ground e,.,g j,7..,g.g.py . y, /g j mocons that rmght be expected if large subduccon earthquakes L.m ne. g 4 occur in the Pacific Northwest. Rey assumed that gap filling carthquake sequences that are similar to those alttady observed in "u ^ m^A
.'"". ,,%. "*7/ *"
5";'.7,g" .,%f,',""'
southem Chile, southwestern Japan, and Colombia may also occur
' % 5. Wnuic dramg d & simulumn d an unhquake sinulu to &
in the Pacific Northwest.De I#Eest e . uakes in these se9uences 22 Alay 1960 Oulcan cuthquake v.i. 9.5) by the suprpouuon of 1978 range in size from Af. 8 to Af. 9.5. Strong monon records are 3gy,g;.ogj earthquakes pt. 7.5).
available for shallow subduction euthquakes u luge as Af. 8.2, but strong ground mocons have not yet been recorded for lager cuthquakes. licaton and Hartzell assumed that pound rnotions seismic moment is at leut 100 times that of the largest enthquake from Af. 8 earthquakes on the Cascadia subduction zone would not for which we have strong motion data (33). lleaton and 11utuu be systematicaUy 6ferent from the motions recorded during Af. 8 (28) simulated the pound motions from giant carthquakes (Af. >
carthquakes on other subduction ones. For canhquAes ofAf. less 8.5) by summing records from smaller carthquAes in such a way than 8.2, their approach is to simply cor.strua suites of pound that they simulate the occurrence of luger carthquakes. That is, the mocons that were recorded under con 6tions similu to those records from smauer euthqudes are used as Green's*funcuons, and cristing at sites for which pound motion estimates are desired. this technique is often referred to as the emptrical Green's functioru lleaton and liartzcU (28) coucaed 56 recordinp of strong technique. A schematic diapam of a model in which the 22 Alay pound motion from 25 shauow subduction carthquAes of Af. a 1960 Chilean earthqude is simulated by a collection of smaller 7.0 for their study. Pseudovelocity response spectra for ground carthquAes, in this case the 12 hiay 1978 Af. 7.5 hiiyagi Oki motions recorded in the 6 stance range from 50 to 100 km are carthquAe, is shown in Fig. 5. In this example, the 1960 Chdean shown in Fig. 3. ney also prepued similu figures for other earthqude is simulated by the superposition of 120 Aliyagi Oki dis ~ ce ranges out to 300 km. One of the most stnking feuures is euthquakes. hiodels of this type can be used to simulate teleseismic t% luge depee of scatter in the spectra for ground motions P waveform data as well as strong pound motioru. Although there
- med at stmilar distances and from similu sized euthquakes. ue no strong motion records from giant earthquakes, there are TD xatter is troublesome when it is necessuy to estimate the records of telescismic P waves (27,28) that are used to constrain the pound motion that a puticulsr site will experience. Even if the modeling parameters used in the empirical Green's function tech.
euthquake magnitude and d stance ue known, the resulting pound nique. The teleseismic data suggest that a luge put of the scismic motions are suU uncertain by a faaor of 10. Aluch of this scatter can energy associated with giant euthquakes is of very long period. This be attnbuted to 6tTerences in the response chnacteristics ofinevid- very long period energy is outside the frequency band of cuthquake ual recor6ng sites (28). Thus, refined estimates of ground motion engineering interest. lieaton and liutzeu (28) concluded that
.hould be obtained by determining the site response from the summing 120 hiiyagi Oki records simulates short. period enerp-pound motions of small earthquakes. from the 1960 Chilean earthqude even though the ratio of the total Another feature of ground motions recorded during luge sub- seismic moments for the two earthquakes is at least 1000:1, duction zone enthquakes is their large size at very large 6 stances. In The results of the empirical Green's function simulations ne Fig. 4, we compue pound mocion levels for Af. 7.9 earthquakes summuned in Fig. 6. These curves represent a best gueu of the pre 6cted from repession ana'ysis of euthquakes in the westem average response spectral levels (5% dunped) for horuontal pound United States (20,30) with those predicted from repession analysis motions obsened at points locned about 50 km irdand from the ofluge subducuon earthquAes in Jcpan (31). At 6 stances of rnore than 50 km, pound motions from luge subduction earthquakes are espected to be fu larger than those from luge crustal enthqudes in ,p, , , ,
the southwestem Un red States. As can be seen in the response /\ N I
%f /4 g/
r! ..
spectra] veloemes at I second, this effect is pronounced at penods of 3/ \ . */ ,
concem to large structures. Unfortunately, the westem Uruted s< ,N
, V A >
States cunes are uncertain since there hae been no strong motion / N / W / \ -
recor6np from earthquakes of this size in the westem Uruted States \/ \/ \ 4 and these nanes ne enrapolations outside the data. Peak pound {0 y\ // , Y '[
sccekrations and velocines for euthquakes of magnitude less than 7 are not dramancalh 6tTerent for Japanese and westem Uruted States g ,',\7 s
/
/ ' <
d
"\' j carthquakes (32)."The onpn of the 6tTerence between pound .
/\ / -
'"\ 'N motico esumates at large 6 stances for large subduaion ernhq~unes t N/ s
~c ,d Y and lup crustal carthqudes in the southwestem United States is lT />
7 \ ,,/ \ us not yet ruuy undentood.
j ,
N e
'/g\//K \ ',Q \
7 N\ ~
Simulating Ground Motions for Giant
, , Earthquakes % i we" i we i ~ Us- i Tie If the Casca6a subduenon zone is strong!Y coupled, earthquakes beunates d the sananon in asuage honamal grmnd emtum l far tuEer than any of the events for which we have stronE motion reslume spectra (5% darnptng) as a Nncuan of energs magmrude for sites records can be postdated What rtught the pound motions look like barcJ 50 km Lnland fnxn the cout $(aner of acual data shiut mean s abes j from a giant cuthquake such as the 1960 Chdean earthqudet its rrm be suru!ar to that in the data obened in bg 4 1 lo APRJL l98* ART 1Ciss 16*
cout and for a vuiety of sizes of subduction earthquakes. Ground b locked, a sequence of seven] great euthquahs (M. 8) or a giant mooon estimates for cuthquakes ofM. < 8 are bued on the direct carthquake (M,9) would be necessary to fill this gap. If great observation of strong rnot%n data, whereas the ground rnocoru for subducnon earthquahs occur, then relatively strong shaking can be luger earthquakes are estimated by the empirical Green's furicoons expected over a luge area of the Pacific Northwest, includmg the
(
s technique. I.arge scatter about these avenges (about a factor of 2) Puget Sound and Wi!!amette Valley repons. Large and potentially -
can be expected for suites of actual data. For the wry largest destructive local tsunanus would be expected if large subduction carthqa, moooru may be about 25% luger at coutal sites and events do occur. Great earthquakes, such as thoec in southwestem about 67% as large at sites in the Puget Sound. De duntion of Japan or southem Otile, have caused great damage over very luge strong shaking during such giant earthquakes is expected to exceed 2 regions. The suEgestion of similar events in the Pacific Northwest is minutes. Avenge peak accelentions may be in the range of 600 disturbtng.
2 cm'sec for coutal sites and 250 cm'sec2 for Puget Sound sites.
1175L8NC53 AND NOTES Tsunami Hazards I I " Dd'am
- 2. C Ndurnura, s"2 3d=^ ' '. Hey, K'"""
MT 89,Cf1 243 (1984)^" " " 74' 0" ' b 5.Wdam,R.N All of the regions that we considered to be potential analogs'for [ [ $ [gg,,'g'j.D u d% CYg,*
7 672 Nik$'"" b the Cucadia subduction zone have experienced luge local nunanus
- 5. L D.*Kulm n a, Ada i(otran Myd y" Regunal Arim smes:
ne 1944 M 8.1 and 1946 M. 8.1 carthquakes in southwestem Japan generated local nunanus that had maximum run-up heights of y yy,,,jQgg/*A,ho'3 h
- 7. R. s cit i
, u.s cat s= or a3-se(19:lig6 sammnwy when 7.5 and 6.0 rn, respectisely, and the 1707 southwestem Japri a h$",g gy'fy'"d "
- g, "R"Mih P"8 5"i earthquake (M. > 8.E jrobably generated an even larger tsunami 9. L Ruf arw H K2namon, re Nis-ia 1.= . 2.1,2s2 (19ao)
(R). The 1906 Colombian earthquake aho generated a luge local Q { y y,eg hp 8g3=d 5,",,d4 74 H3 0'84b73 tsunami that extensively damaged much of the coastal reg ons of 12. H Kmarmn ard L Asas, Emy ar r,w.a Am 3,30s (19sst southwestem Colombia and northem Ecuador (R). The 1960 13 I C 5e*(e, M tm.,as, w. H. Pre. ant,t c.,* A.. s6,4929 (19:1).
Oulcan earthquake generated one of the largest tsunanus in recent 14. .Man,urnro Im.ia; ard io te J. C seE, unpitenhed o k, manusmp Trtexh pardel stram rues aorg cuennamal uran akes the aus a mammum times,yith heag damage occurring both locally (35) and abo in Hawau and Japan. Although the inmumum local run up may have g $N" , g p,, e ,p i, 7g exceeded a heignt of 20 m, it seems likely that most of the coutal la st L Redinger ard J. Mams, c.yh An tsw 9,401 09:2h regions adjacent to the earthquake expen.erv.ed run up heigho of is. 17. G riaAcr ard 1001 09'Ob c n,3,,, , n,i Sage, c., af God a., se,r Am ,fA.# II, f m, t g3,,ig g, g4 (3,,,,i less than 10 m. If very luge subducuon euthquakes do occur in the Amsemv cd sacrxa, w 19 %
Pacific Northwest, they will altnost certainly be accompanied x gby $ I iS'*d' 'j',3 "y' O ""*"', ,,,
g,,q,,,,, f,,,,,,,g , 3 tsunanus. It is di6 cult at this point tc, re!iably estimate the tsunami nr 54-wan,v.o 2Y b,lapu 19ssn KTta,oie,u.s c.s s.,,7 run up heights that may foDow any large euthquake on the thAia 21. N fr[ r,a3v 3 g,lM subducuon zone. However, run-up heights ranging from seura! Ar. ster, e op., 7 A a .Ja.109s6), p.130, 8 F.
meters to sesen! tens of rneten have been observed along other 21 i Mams, r peparnanry=nais a e %wn cmi- m vaj subducting boundanes after events of the type that we consider to 23. [Gnggs s'rmi m, c.t ser As A.2 81,13?o 09'Ob .
be feuible for the Cascadia subduction zone, 24. T. H. Henm and r. D snnete, Jr , a.a. Sr d sn A. 7s,145s (19ss). !
2s. M Ardo,7 24,119 0 97s). t 26 H Karwrun L C Mda!hr, A.A Sr.ind ser Am 72,1241 0982). i
- 27. s H. Harric11 and T. EL Hesam, sh4 75,965 0985). I 24 T. H Hea m and s. H. Hartzca UJ cast sert (yr. F4 A, Ao.32J 09se). '
(Onchl3 ions 29. W s. }c=mer ard D M smee, s.: sev=s sn A = 71,2011 0 981).
30 uJ c.s 5- opr.Je A, 41977 09:2r Strong evidence exists for acti.~ convergence at about 4 cm' year 31. K. Kansahama, K. Amata., K. TakA%anha, m trurasher af de ty6:4 M'au c,.tr a +1mirrerne Hamsw cui, NJ,19u n 1 on the 1200 km Cecadia subduction zone. Furthermore, the vof2 23 7. l phvtical chnactenstics of the Cucada subduction zone resemble l[ k Op*' "'f, *2N!
those of other sutduction zones that hue expenenced luge shallow 34 r. A laten e ard
/g%'k. H.[97ps, smah, . r,
, s oner er r.4 sam. (my) (Netmal j
N" ***"" d ^"*"F # " Ad""""'"" *- {
carthquales. Even though there have not been luge hutonc subdue. Mc'$,D',uj'""6 i l tion euthquakes in the PactSc Northwest for at leut 150 yean, the as.Htsem,ov9^I o narrr., s.: s-d s. A. 53, litsW-0963)
Caxadia subduction zone ma7 be storinE stnin CU#III to be "'^'" " 8 ' h F ""'" d '" #d *' " 'h'
re! cued in future great earthquakes. If the Caxadia subduction zone 36hardier, )
"' F*'^.'s*e i
'h'^ihr manunnp ge,8 Armater,'an i j
\
168 SCIENCE, voL a)6
. . . . . . . . . . - - . . - - - . - -- . - . . _ _ . . . e , . .