ML20151H144
| ML20151H144 | |
| Person / Time | |
|---|---|
| Site: | Trojan File:Portland General Electric icon.png |
| Issue date: | 04/12/1988 |
| From: | Crouse C, Schell B, Vyas Y AFFILIATION NOT ASSIGNED |
| To: | |
| Shared Package | |
| ML20151H012 | List:
|
| References | |
| NUDOCS 8808010159 | |
| Download: ML20151H144 (25) | |
Text
-
( Bulletin of the Seismological Society of America Vol.78 February 19S8 No.1 GROUND MOTIONS FROM SUBDUCTION ZONE EARTHQUAKES BY C. B. Cnoest, YocesH K. VYAS, AND BRcCE A. SCHEl.L ~~
AssTRAct A total of 284 horisontal components of acceleregram data recorded at sed setos dunng thrust, normat, and stnke.sl65 earthquakes occurring in seven sub-ducten sones around the Pactfic Ocean were analysed. The results of statistical analyses of 5 per cent damped pseudo velocities (PSV), computed from these accelerograms at 10 periods (T) between 0.1 and 4 sec, indicated no significant 46tterences in the average PSV levels of the thrust, normal, and stnke slip data from the Northem Nonshu aone. Each of these groups of Northem Hertshu data, which tegemer cornprised one hal* of the total data base, were fit equally weit by the same attenuation model. The analyses also revealed met at intermediate penods (0.8 3 T < 3 sec), the PSV observed at saff tod sites withm the Northem Monshu, Nankai, Kunt, Mexico, and Alaska zones were, on tne average, sigruft.
candy greater than the PSV observed at similar sites in the Peru /Northem Chile end New Sntam/tougainville aones. This observation indicates mat differences in m. som. and/or iravei pom characteristics betw.en the two groups se.
counted for the differences in PSV. Independent evidence supportmg soutco l differences is the correlation noted between PSV in mis pened band and the charecteftsbc source completities for these tones, which Hartaett and Heaton inferred from the teleseismic data of the larger magnatude earthquakes. Certain anomalous tectonic charactenstics of the New Bntain/Bougainville tone were noted that may have contnbuted in some systemabc manner to the relatively unusual spectral characteristacs of the ground motions recorded on stiff sod sites withen mis sone.
Some of the differences in the PSV among the aones at intermediate and long penods were probably the result of differences in local geolog6c charactenstics.
Geology greatty influenced the Monican PSV data from soft sod sites. To a lesser entent, geology probably affected the Alashan data, which were recorded mostly on deep soft sode, and the New entain/Bougamydie data recorded et Yonkl.
which is undertain by sener sea depoons.
INTaoDUCT1oN The charseterittie: of pwnd motions generated during subduction. cone eanh-quakes is a subject that has received little attention in the United States relative to studies of ground motions generated by transform margin earthquakes in the Western United States and intraplate eenhquakes in the Eastern United States.
However, the recent suggestion of the possibility of a large subduction rene earth-quake in the Pacifle Northwest (Heaton and Kanameri,1950 and the well-documented occunences of such eanhquakes in Alaska and other regions of the world indicate the need for a better understanding of the ground motions from these eanhquakes, A considerable number of seis:notectonic stud.ies :nany of which are referenced in the nest section, have been conducted for subduction roces around the Pseule 1
8808010159 DR 8806is ADOCK 05000344 PDC
s 1
1
(
2 C. B. CROUSE YOCESH K. VYAS, AND BRUCE A. SCHELL rim. Recently, sufficient pound motion data became available to permit a syste=-
atic study and comparison of the pound motions among these subduction zones, For this study, processed and unprocessed acceleropsms or their response spectra were compiled from various sources (Denham and Small,1971; Prince it al,1976; Brady and Pere:,1977: Rascon et al.1977; Espinosa et al 1978; Crouse et al,1950:
h!ori and Crouse,1951; Beavan and Jacob,1954; Silverstein et al,1956: Anderson i
et al,1987; the Japan Port and Harbour Research Institute; and the U.S. National Oceanic and Atmospherie Administration). The unprocessed acceleropaca were corrected using methoda similar or identical to those described in Trifunac and Lee (1973). The 5 per cent darnped pseudo velocities (PSV), which were computed from
- the hori
- ental components of the corrected acceleropams at 10 periods between 0.1 and 4.0 see, represented the ground motion data base that was analyzed. Some 1
acceleropams, recorded during the larger subduction rone earthquakes, and their assxisted velocity time histories have been examinet. m a parallel study by Hart: ell and Heaton (1955b) and Heaton and Hart: ell (1956). They identified similanties between these strong motion records and the source time functions derived from teleseismic data recorded it. Pasadena during these events (Hart: ell and Heaton, 1955a).
In this investigation, statistical studies of the PSV data, using repession analyses f and analysis of covariance techniques, were cendreted to determined 1) the potential effect of focal mechanism type on pound : notion and (2) those subduction tenes in which the recorded ground motions were similar or significantly different. Possible i reasons for these differences were explored by examining the regional and local
- geology and the seismotectonic charseteristics of the subduction zones.
DATA BASE i
Sesimotte:cnie characteristics of the subduction :ones. The locations of the seven subduction zones considered in this study are shown as wide lines in Figure 1. '
Certain seismologic, geolegie, and tectonic characteristics of these :enes are sum-
' mari:ed in Table 1. The information comprising this table was compiled from many references; the principal ones are listed at the end of the table. h!ost of the column ,
headings are self explanatory, except possibly column 14 PQ structure") and column -
i 17 ("Seismic Slip'). The designation Low Q" refers to tones of extremely high attenuation beneath the high plateau of Per.. beneath the Sea of Japan, and beneath the Sea of Okhotak (Kuril zone). The other renes in these and other subduction zones have relatively low attenuation properties and h:ve b, ten des 4-nated as Normar' in the table. However, it should be noted that little is known '
i about the Q structure from 0 to 150 km beneath the Earth's surface, the depth
' range of nearly all of the earthquakes in this study. Because the paths of the seismic waves travel through this portion of the lithosphere and upper mantle before arriving at the recording stations, detailed infor=ation on the Q structure at these shallow depths eventus!!y will be needed to better interpret the recorded pound motions.
Nonetheless, the Q structure, as it is presently known, is provided in Table 1 for completeness. The values in the seismic slip' column are the percentages of plate.
convergence rates that are released as slip during major earthquakes.
The categories in Table 1 were selected because they are commonly discussed in i the literature as hasing some bearing on the : node and mechanism of the subducti;n process or as being a direct result of the subduction. Some or all of this information may be indirectly related to the earthquake source and travel path character:stws, j I which in turn wou!d affect the characteristics of the reecrded pound motions. This
)
on0UND MOTIONS FROM SURDUCTION. ZONE EARTMQUAXEs 3
(- .... .. .
,J. \r
.. ( /' t, W \/.i .
(
b.
p 98 %
mese.v t~o N w +e t*
) . =e u n.E N **" /[
l [4 h ==
s a n's -
7)
D % *\.
.n.
. . . ..**i, dr %nte:=n 1
7/y N' r"
- n. 3w',, . f ee.Q-W
" ( ,m, l ,4 "
i .
.es . g., ,
y 6k 5 , ,;, 7 -
~
..u
~ .
,; %" ( ,
) ..*J
) L.
~c .- s. ,,- .... u. u. ... ~*
Fic.1. Lacanen map of subductica senes around the Pactae Ocean. Solid tien represent subes.;en tones math med-deveicped seatloor trenenes; d. ashedd ' nes edacate subducuen tenes witnout me.l.
deulo;*d trenches or indacate other 14thosphene plate margins. The lateied made 604.4 lites and atde daane4 Imes are the sones consu:lered la this study possibility was explored to a limited extent in this study by examining possible correlations between the seismotectonie data and the ground motion data.
Acceleregram date. The relevant information on the recording conditions of the
' acceleropam data base compiled for this study is given in Table 2. For all but two of the entries, the PSV from both horizontal components of the corresponding accelerepams were used in the subsequent analyses. The exceptions were the -
acceleropams recorded at Pajaritos, Mexico, and Santiago, Chile, for which only one horizontal con:ponent was available. The total of 255 coc:ponents indicated in Table 2 were distnbuted among the various subduction zones according to twe of focal mechanism 1 thrust, normal, and strike slipi as follows No. of Compcnents su % en:n. The.n smai sree.s3 Toes Nortaern Honshu 94 :: 4 1:1 Nu.'a u 22 10 6 3:
Ltd is Alaaka s s 4 ::
4 -
10 Peru / Northern cme 13 6 -
Mese 19 19 4 -
23 New Batain,Bouetaavule 12 4 -
16 Total 182 4 15 15 All earthquakes in the data base occurred within or near the Benioff.Wadati zene of the subduction.2one regions. Focal mechanisms were cornpiled for these earth.
quakes from ths appropriate references listed at the end of Table 0. Although most of the ese.hquakes were thrust events, a sigr.ificant number were normal. As expected, relatisely few were strike. slip events. All peat esithquakes 1.\f 310) in
i i
1
( 4
- c. a. CROUSE. YOCESH M. YYAS, AND BRUCE A. SCHEU.
l 8 8
- 12 t- - 8 8 8 If f f
- ! i, ! !.I g1'l1 'I,- i l
m!ir}ir 'i
.t k
t, _14
_t .t.
1
" 1 l
i.,h k k h
h h ieA J' !
i g g i ,]
h
, k s k s ,
i 2
4 f,j'l
] ill '
A 2.
A ill] i]ill! !0l, ,.{ (jl=,lI;{ 3 o
i !
A A 1 -
L .1
, i- I ! ! ! f J I 3]j i 1 t
' I' j I .t ....
j
] = fc ... . . . .. . di...4 di i] j J,Ig ;
i 1L idi di [
]
i i
i di di sw 4.1 1 l- =.
E]} _. .
l d
i : -]l A f S .E !
i d
.1
.I" t
i ! I A 3 !! ! ! { -
s, r; l., .
- * - ); ;I g ll ]$ : . }<
1 f'
ei
.= a- J - t -l
+ -]
i i
3 [ $ .I =a
- n 8
f J/. -[ .i {. $';, $ [ lA i 3
1 E# i,. i i e *
- i a l-i E <!
s 0. 31 1
e'? (! ;i l _. =
.' I L:j -l d :y* h.] ))
yli ej '
i 1 'O 2 J J J g
, 8 s J J J
';- ' s: 1 i . :. I I
! 5: , I' l' l' l t' :(J.g{.y i4 i
3 .1 I
[ j. l'l.3 j J jkJ j J j'k' j',J bg}' 3);j$.!!
l j,, r
!I I ! ! ! ,5 jl h)$1 gd .
l} l
'l {g{~ji'j
! 2 -
K 1 5 e :
1 g i. ni 'j . '
l* E,.*l!8'A !!! *I* A!! 3}
<- g i;5 E
, $18 $!* ,5 )p, i I e
! l- 'j *l -* - : EJ 3 11*i2!.,},a 3' j I
i 1 I!t 1 Jj i e ll.i i ;
t
- ., 'l-Het 2
- > 3 5 4
' ji" j {t. :.1 .[ E -
-l w = - js; 4. lc! l 3 v -
u , v F v 3, e
l - [ , [. . 8'****
Norshem Homhu, -
Japaa 54/0:/14 4.0 N 43 l 21 50 54/02/14 60 N TX0:4 t 45 23 51 62/04/30 4.5 T TX0?4 :
35 64 49
) 43/0$/04 41 T TH001 40 53 63 43/03/04 KT001 4.1 T 40 $3 63 63/04/04 KT003 ;
5.1 T 39 35 5:
64 % %5 40 KT014 7 34 34 65 64/02.95 40 7 KTool.a i 54 34 65 64r02.t3 4.0 T KT001 b ,
$4 34 45
] 44/11/14 51 N KTool e 63 5 69 l 4*/11/19 60 T KTX1 45 SO (9 KTN1
". 61 05/16 6: T 00 90 1*9 1 M C3/16 6.: HK001 T 00 319 I 66/05/16 4:
I?O HKX4 t T 20 !!9 54 64 05/16 6: T HKn3 ,
' 00 243 113 M $5/16 4.2 T THOM M 167 ?1 TH3-3 i
M/0$i16 75 N 06 034 '
64/05 16 75 N 115 HKW M 07.01 06 in M NKJ13 ;
4 4.1 T M 54 64 07J05 44 91 TK:'Ls 7 44 6: ?6 M /10 '06 53 TH0c3 T T1 15 70<01J01 6? T 12 KTT4 .
25 46 31 HKW !
71 06/13 53 T e ?!,0AC: 55 M $5 KTxl
? .0 N 45 196 201 71/10'11 52 T O HKuM
- 2 00.'09 1: 4: KTSS)
?.1 T $0
'3/11,19 25) 240 KT%4 44 T 54 107 l
til TH033 6
?4 @ 03 4.1 7 49 i
74,t*. 04 39 63 KT014 1 43 T 45 !3 v 09 04 l? KTM4 16 i
, $ !! 40 4? TH209 i 74/11 C9 63 T 105 15 124
} '4/11/16 4.1 7 44 HKM6 16 56 j *l C4 1: *? T KT334 40 114 10 J ?6.M 1: *7 T .0 THMS 100 10: THOM i
M-0514 la T M i 254 ;!) H Ky,4 M *4 1 ?: T M u4?44 li) 19: TH0;$
s: s 1x to :: '
i M49 21 45 T KT>x ,
ST 194 70 01 21 6*
N4 TH;;)
T '5 10:
i 7044-01 54 T ?$
1:5 HKN 101 1:6 TH0:9 1 ?! 44 #.' *0 N AS Of2 7:m.a 264 HKX3 a
?:42 09 7.1 T n :u as c a ..s
, ?1 T M K4
?:e.n X4 TX054
?* 03 N 64 s n x4 ni Hxm i
- 6. 4 5 50 55 97
}
i ?: 1244 ?. T TH009 ,
64 213 (1 a ?:1 44 **
T +4 KTX4
?41147 40: 407 1 63 T 1:5 Oeanams.m s 229 031 HK3M 1 ?510 M 40 7 60
- ?l-C 20 4.?
174 34 HKW '
N 50 000 235 j *540 0 4? N W Onasma.fi i C4
?60612 **
T 40 011 inasta.: !
j I?3 1*6 Ous4ma.y $ I
?& '4 12
- ? T 40 4)
! ?l 4 12 *? ? +0 23 : Aiaa.5 f
- 3 ;?$
1I ?i 4 1: ?* T 4 3?3 TH209 !
JT5 KTx4 i
3 1 I i
l l
l
(
6 C. 5. CROtl$E. YOCESH K. VYAS, AND BRl!CE A. SCHEL1.
I TABLE :-Coesed 1 - c
. ,, :'.,, u o . :., L :., >~
n.0i.n 7.1 N 13o ::3 :d TH0:e 8141 03 ?.1 N 130 103 166 HK003 42 03 21 ?! T 40 137 143 HK0c3 82 4 :1 f.1 7 40 64 75 Tokschi41 4:47 23 70 T 30 111 !!: Ka Mme Z4kan.$
$24**3 *0 7 30 00? 009 TK056 63 05 26 *$ T le 107 104 u 't $
&30526 ?.9 T 14 1*3 174 $aista $
$3426 T.S T 14 .44 .44 TH009 Nanaat Jann 65,04 70 61 $ 40 21 45 CB002 6544 ?J 6.1 $ 40 30 50 CBM 4 11/1: 53 $ 20 19 :S KSNS 64,04/01 *S T 37 60 ?S K5W 64 04J01 7.$ T 3? 1:3 135 KSX3 M<04 01 f3 T 37 167 166 $ K003 64 04/01 *3. T 37 60 M $K:06 66 06.06 66 N 48 116 14 CGM 5
&l 04'06 66 N 46 14 ?4 SKX4 49 % :1 65 T 39 43 to R$N2
?O'0716 6? T 47 *1 31 K5N2 70'0?/06 &? T 47 ?O S4 KSN3
, 70 07 06 61 T 47 21 31 KiOC2
( 71.01,04 61 N 44 64 *4 KK006
- 3'04 21 6.4 N 1: 41 37 K5014
- 4 4 09 49 N 10 5: 53 $btrasw.iet. tan $
?! 4 :3 6? T to 141 213 Sto anate N!
!s 0?.04 6: T 1:0 ;;6 !6 $KMS
?t4?44 62 T 100 &? 146 N!aassaph!
Kani. Jann 4: f4 :3 ?O N M 71 1CO HKX4 MJ10< 6 ?.1 T 15) 16: :0? HKX4 4 0407 3? T 4 2: *1 HKX4 i
?; c.411 51 T 63 33 it H K X.4
?3 M !? ?$ T 41 11: 134 HKW4
$d 4 19 ?. ! N 06 031 33: HKX4 041 4: T 43 :l? .% HKX4
?3 4 :4 ?I T 30 194 lif HKX4
?l-01 ;5 ?6 T 40 44* 464 HKX4 71 10-M ?? $ ICO $4 X1 HKX4
?812.C4 *? $ 100 401 41: Tasub4!
Ajaaka *1,01 01 4.0 N M 41 100 :*00 i Axura ge i
- 3 01 01 60 N M l: IN *031 Ancherage >
l ?S 01 01 60 N M *) M
- 4 ' A nt 7, e: .:. ?6 T ti > a n: = uragei .e
- 9 0 ?$ !4 T 13 164 1:? :*;l iY a.a wt at i Pera N CLe M 10 I? $1 T 38 :4 ::9 4Xr Lea ,
?00531 60 N 43 3?: XC 4*ct I- ta i 71 11 Oi 53 ? 5* 1;* 111 40,00 L:t e i 74 M. 01 46 N H *4 l'3 430? t ca.
?40105 44 N H *3 13 4A.:3 Octa.
- 41003 61 T 13 W "
42 L eai
?4,10 03 41 T 13 91 . 4m k
'411 OW *:
T 4 H w
- 4 11 C9 T 4 13
- 10? T9 ?5 T !) 14) *
- n. m_. = n
CROUND MOTIONS TROM SUBDUCTION ZONE EARTHQUAKES 7 TAllt.E -Connm.ed 8d w a3d* ,,,I.,i 4 i [., [., [., 8" f
Mo nico 62/05/11 7.0 7 37 ;60 263 + Alameda Central.
D.T.
4 '05/19 7.0 T 19 060 Ott + Alameda Centrat l D T.
64/07,06 4.7 N 100 240 260 l
+E. M. Gontales. D T.
64/0?/04 4f N 100 240 260 *P E. Hidsin. DI.
?S/03/14 58 N 155 40 ifo M 9tT.Gwt>4rnsi }
?4/11s09 ?7 T tl OH 300 +6316 iMinatitMrd 74 /11,T9 f.!
T 14 329 330 +4300 ipa:entm it '11i:9 T !$ 119 100 M19 iOaxaca e '
$5 09.19 84 T 05 4f9 40 $isme Pueb:a M4919 $0 T ;$ 3M 3M Tacuta3 a. D T. L M 001e $0 T $ Ji1 141 -sistne Weree. D F.
65 09 19 90 7 $ 15# M9 -C ae Atsetos F D T.
60 15 09 19 T $ 3M 365 +5ec. de Coram . D F.
Ne* Br.tatn. Bowgain. 70,10/31 ?.0 7
(
45 154 140 -Yonti. UpNr Rams l de *0/11.1 45 T : !!) 141 -Yonki. Uppt Ramu 71,0?!!4 10 T 43 000
[
205 Perar.a l 71.0? a6 S.1 T 43 096 301 Panter.a ?
?! 09 05 !) N 111 43 119 1.aeBan *
?1c10 ':5 65 T 107 250 27: Raba21
?: 01.19 64 N 100 2o3 206 *Venkt. UPNr Rm u 03,1$ ?.i T 9 :?! Oih Purana
.Wres '
V = magstuce iM., AI or M. s; T = fault typ iT = threat. N = acreak 5 = stnie shoc
% = fxa) de;th. a = epicentrai distano: R = center.of energy.n;nn d. stance. $ tat:en na=n pre <eed by a p;ss denote sites on acft sc1 Data from the avbdacton scres .ere >bts.ned from Anders:n et 3L dMf). Beavan and Jactb i!N1. Boatur.snt (19Mi. Brady and Pere: 49'*1. 3 era of t*.* Inte wmai Se.s we. cal Cewe, Chael and $tes an dH ). Croa* et a; 49Ms. Dennam 49??n, Dennam and $ =a.1 41971h. Desty ashi $Nace i!9?el. O. Dunghv 'Mrknal eccmunicatson.19W Dr.ewor.sta ana Wx4 house 49% Da.coonsaa et a; d)Mt. Espinosa et al 49?ll. Tw.nta and Kanamon QH11. H4.9te.1 ar.$
Heuen $1h5ai. lca.ha=a 49?D. Jacob and Ha.aston 4t01. Kanamen ell?D. Kasar. ara and insta dH4i. Lafnn ud McNady illm K. McCee <NrMaal ccumuracat4en.19m Molnar and irku GH6h Man and Crowe dHD. Pnto et al 49?41. Rasun et a 09??L ieemeen er at itW . us itaucer and Wakha dl?4L Ter the thrn Jaur.ese subdact.cn scnes. aates taa.uted =>ta skaan i i reprnent tround.su6sn data frem N:n and Crwae GHIL Dates inicated wah asaan e 6 nNuer.:
tata egewed 4a 1H4 by the .'apan Pen ud Haitious F4 parch Inn;t.te ar.4 Ann Man af K.aostan Comastanti.1.11 The ! cut-d.pt asetert aes. pat.tg the Alaskaa. Perman, and Me a;can 6 tat 4ns tort an nu by ide v.3 coepea; seer is. ant er a;. Inn.
the data base, except one, were thmat events. The notable exception is the 31.\tay 1970. Peruvian earthquake Of, = LO), a normal faulting event which occurred near the Peru trench.
Note that approximately half of the components were recorded dannt earth-quakes occurnnt in the subduction zone adjacent to the northern pornon of Honshu Islard. Japan. The number of components from the other zones is much smaller, and questicas regarding the adequacy of the data sample size are cer?Ainly relevant, For every one, the sample site could have been enlarpd by sdmitting acceleret: ms that, were recorded. for example, at sites on geo!cr.e media other than soil. Had rock site record.s been considered, the Alaskan data base wou d have ir:reased substantially, but the data base for the other genes sculd have only slightly l increased. Thus, a deculon was made not to use rc<k site data in order to remove one variable from the analysis. Ancther example of acceleregrams that were net
t
.. .t 9
(
l J
i S
C. 3. CROtl3E, YOGESH K. VYA$, AND BRtJCE A. SCHELL 4 selected were those generated by eanhquakes dennitely not assoelated with a Benioff
' none, or earthquakes with mapitudes and epicentral distances well outside the range of most of the selected data. Partly for this latter reason and because of the small sample sire, the eight horizontal components of the acceleropams recorded 4
- at Olympia arJ Seattle, Washington, during tne 1949 and 1965 esnhquakes, which !
I occurred within the Juan de Fuca subduction zone, mere not included in this study.
Accelerogram data recorded by the S5!A 1 accelerograph network (Chang,19s4: i Chang and Chiu,1984) and the Sh! ART 1 an.*y 1 Bolt er of.,1952)in Taiwan were also not included primarily becaun the causative eanhquakes occurred in a region *
- where the tectonies are complex and include both transform faulting and subduction.
Thus, most of the eanhquakes could not clearly be asacciated with etther t!.e transform faulting or the Benioff Wadati rones. ;
l Accelerogram data recorded during the 3 >! arch 1955, Chi!e, eanhquake (Ms = '
i 7.5: EERI,1955), which would have greatly enhanced the data base from the Peru / '
Northern Chile subduction tone, were not available for this study.
The accelerogram data used in this study were generally recorded on shal!cw, i
- stiff soil and sedimentary deposits between about 5 to 25 m deep over Tertiary or i older bedrock. The notable exceptions are sorne of *he data from >!exico, which >
were recorded on deeper, much softer alluvial and sedimentary deposits in a basinal structural configuration Other stations underlain by relatively soft and/or deep soil deposits or sediments include all the Alaskan stations and the station at Yonki.
Upper Ramu. in New Britaln/Bougainville. The stations at Anchorage. Alaska. are j i underlain by stiff alluvial material on the order of 100 m deep over Tertiary bedrxk
- iSchmoll and Barnwell,1954; Idriss.1985: D. Cole. personal communication.1957 L j
1 Both the Yakutat and Iey Bay stations in southeast Alaska are underlain by ;
generally stiff sediments, which are more than 200 m thick at Yakutat and probably I much thicker at ley Bay tYehle.1975; G. Plafxer and B. 5!cinia. perscnal commu-j 4
nications.1957L The Yonki station is underlain by 110 m cf poorly consolidated (
l 1 and unconsolidated lake sed!=ents IK. S!cCue, personal communication.1956L The potential effects from all of the above.noted differences in !c<a1 geclery were considered in the interpretation of the resulta cf the data analysea. l 2
The moment mapitude was used when known, which was generally the case for l 4
the larger eenhquakes greater than mapitude 7.5. The moment magnitude scale i !
better represents the true size of great earthquakes, and its use in this study was i
j preferred. When =oment magnitudes were not available, surface wave or Japan i
- 5!eteorological Agency magnitudes were und. These magnitudes were used fer the !
I smaller earthquakes approtimately less than magnitude 7.5. They were considered '
to be equivalent to the moment mornitudes based on Utsu (1952) and Heaton er al I (1956L The center of. energy. release distance was def!ned as the distance from the i) reecrding station to a point on the fault rupture where the energy was considered ;
to be concentrated. For all eenhquakes less than =apitude 7.5. this point was i i assumed to be the hypocenter. For =est of the larger events, this point was the ;
centroid of the fault plane deilned by the aftershe<ks. If special studies of source charactenstics and aftershxk distnbutions were both availabie, the center.of-energy release was ulected at a point on the fault plane corresponding to the i locatton of greatest energy release. An example of the latter is the 16 Stay 19f s, !
Tokachi Oki eanhquake off the eastern coast ef Northern Honshu. Nagamune (1969> and Fukao and Futurnoto (1975) determined that the scurce of major energy j relesse for this cent was lxated along the mestern edge of the fault plane. This 4
1
i i i t
t cRot:ND MOTIONS FROM StlBDUCTION. ZONE EARTHQtlAKIS 9 4
point was a sipificant distance from ehher the centroid of the fault plane or the epic,m:er, which was located near the eastern edge of the fault plane.
The distribution of(1) mapitudes versus center of. energy. release distances. (2)
I magnitudes versus focal depth, and (3) focal depths versus center.of energy release distances for the 125. component data base selected for Northern Honshu is por. ,
trayed in Figure 2. The three different symbols in each figure represent the three taes of focal mechanlims. The data cover a magnitude range from 5.1 to 3.2. a fxal depth range from 14 to 130 km. and a distance range from 42 to 407 km. The data are fairly well distributed over these ranges: thus, no strong correlation exists among the three pairs of parameters.
3 4
Generally, the data from the other subduction renes are distributed in similar fashions as shown in Figure 3. !n these figures different symbols designate subdue.
tion tones. Figures 2 and 3 show that the magnitudes and distances of the smaller l data sets are generally within the magnitude and distance ranges of the Northern
- u. . ... I
... . ..
- s' l' F3 9 !
, i . ... .. .
l
- 3 9 i,
g . . . . ' .' s e'*."" t I..
l.. I.. '
] . .
, , . l,. ,
. , g... .. >
- ',A '*' *. .. . ". '
l .' ' .
I i , . >. , .
I
. .. ,'.4 1 , .e ' l' #, , '
',s, '
j ' "
" J .. .". * * '
I
..."w. ..
... 6 . ..
"' i Tic :. Dutr%uon af earthpake data in Tace 2 fer Northern Hoesha sure.stica ast.e. Thr.st = f O actmal = a strthe sep = O sER = center.et.emergy reiesse. -
< t i {
.. . ve r
. .. l
- . ,e s .
.' e, 9..',i's..
. '!# g.. , .,, i.
,..'>e 8
{
1
- .. i .
l..., . . .
... t' '
, " *.'. I
- 3,
- 8
... . t, 8
- t ...
1 i .
J . ". * * .' ,
- l a. i " m .. " '
- r. : 6 ..
nc 3. outrwuw er euthpne data .a Tase : for other ,me.ca me. Nein . . x.r.l . .
Z.energy of Maasa re' esse.
=a Fers Scritern We = 0 Mesa = 0 Se= Bntu.BNuade = . CIR = ceriter. I 1
10
- c. a. CRotst, YOW n 4. YYAS. AND BRtlCE A. SCHEtt.
Honshu data. The data for all subduction zones were selected in this manner to avoid excessive extrapolations of the repession equations that were used to compare the pound motions from the subduction renes.
ANALYSES Statistical analyses were conducted on the PSV data collected for the seven subduction zones. The main objectives of the analyses were twofold:(1) to determine whether differences 's PSV within a particular subduction rene for a given mapi.
tude ano distance were due to differences in the type of focal mechanism, and (2) to determine th* vubduction renes in which the recorded ground motions could be considered simil. t. For zones which had sipineantly different pound motions, poinble physical reasons for the differences mere considered.
The large number of components (10s) for the Northern Honshu sone suggested that this data base 3rst be used to determine the possib!e effect of focal. mechanism t>Te, and then used as the basis for comparisons with data from each of the other ranes. Standard ana:ysis of covariance techniques (e.g., Dixon and .\!assey,1950:
Dixon.19sM were used to test whether a linear attenuation model St troups of data le g. groups representing different focal mechanism types or poups representing different renes) equally well at some conddence level Because the analysis of covariance in these references is based on lit: tar tuodels, preliminary analysis was onducted to determine whether such a model produced a reasonable fit to the data.
( Seicenen of a ter.,anen model As a nrst step. all of the PSV data for the Northern Honshu :ene were east into a repession analysis and the coef6cients a. 6. c. and d in the equation in[PSV47)] = a - 6.\f - e In(R) + dh <!)
were computed at each of 10 perieda between 0.1 and 4 sec. In equation (1), PS W T) is the 5 per cent damped PSV in centimeters'second at period T. .\f is earthquake magnitude. R is center-of. energy release distance in kilcmeters. and h is focal depth in kilometers. The values of the coeiScients and the standard errors at each penod are riven in Table 3. Other trial restessions, assuming different functienal forms for the PSV dependence on the mapitude and distance parameters, were conducted.
These equatiens included a quadratic map 2tude , term, e.\f 8 and an anelastic TABLE 3 Rtsttis of RtcREssio% Autists o% 125 CoWPo% TNT D474 B4sts raow rHg Notturns Hossue 5t setettow Zost f5tt Ermos ild p,w ane== Ce areew
- 48 ' st 01 1 s6 04 o2
- 1 22 00&i) 1.00 0 i<i.4 3.19 0 44 -0 is O cC53 3.94 0 (?:
04 1J$ 0 (4 -0 14 O6 00041 0 SO O !)?
) f? ? 15 -0 95 0o??0 0s -0 06 0 *1 o 674 0N -0 6? 300:*
tJ - 1. ;3 L:4
$ 44 )?n
-> s3 0MO o (0 15 -2.?) 0.?t 3 lo -; S4 1.1 s ~)() ->cc0! 0 19 0 M3 L26 -c *5 -0 E4 00; O *11 30 - 1 46 L34 - 0 15 -0 y>46
( 40 -409 t 29 -0 15 0 s? 0.?:0
-0 CC33 0 72 0.? 0
- A.% al a pr-t4 Sty H hn: thesis, N. 4 is sit?.icar.th Li!ertst frers tm.
~
! j i
( i GROUND MOTIONS FROM SUBDt*CTION ZONE EARTHQUAKES 11 attenuation term, /Rt however, these terms, although -hysically plausible were not i i supported by the data according to the : test and were omitted. A repession was also performed on a nonlinear variation et equation (1), in which the term, R =
Cserp(CsM), replaced R in the equation. Such a magnitude dependent term in the 4
geometric spreading term, e In(R + Cierp(CsM)), has been used by others (e g4 OASES,1978; Campbeu,1951; Hadley et st,1982) to account for the saturation of l l
pound. met on amplitudes at short distances, R. However, the standard errors of -- "l(
this more complicated model were similar to those obtained from equation til. ;
Funhermore, the residuals (i.e., differences between computed and obsened values !
ofIn{PSVt T))) were uniformly distnbuted abe9: their mean of zero throughout the !
range of each independent variable. These results suggested that near.neld satura.
tion effects were not signineant for the Nonhern Honthu data. If data had been l i
available at R < 40 km, such effects might have been observed: however, because l i there was no empiries! basis for this effeet, the nonlinear form of equation (1) was j not considered further.
L The purpose of the term, dh,in equation (1)is to account for any possible effect
! of focal depth. The term has some physical basis in that deeper earthquakes might be expnted to produce pester short period body wave motions than shallower !
! r i
ennhquakes of the same magnitude and hypocentral distance because the anelastic attenuation would likely be less and the stress drop would conceivably be peater dicGarr,1954). On the other hand, shallow earthquakes tend to generate more f
- t 1
long.peried surfi.ce waves. The decreasing trend of the d values with increastng
,' period in Table 3 tend to support these postulations. The inclusion of the dh term is supponed to some extent by the data as indicated in Table 3. which also shows !
I the probability that the coefncient d is signiceantly different from rero. These i
statistical tests suggest that d is signincant at short and !cng periods. The insignif-1 icance of the eeefScient at intermediate peric<'s between 0.6 and 2.0 see to espected i
] because the values of d are c!cser to zero In this perad range, d is in transitien 3
from the relatively larp positise values at short periods to the relatisely arge negatise values at long periods. Howeur, it should be noted that. at each pened. i the standard errors from repessions without the depth term are very simtlar to those in Table 3. indicating that the inclusion of the depth term dc4: act substan, tially improve the St of the tecdel to the data. Regardess, linear medels with the i i
{
depth term [ equation (1)] and without it were both used to test (in the pctential ;
] effects of focal mechanism tne on the PSV and (2) whetner there were differences '
' in the PSV among the subduction zones Because the results using both sets of i
i attenuation in this paper.
equations are similar, only the results based on equation (1) are reported E#ect s//x26mechonum tve. The analysis of covariance tests w hether the mean values of the In{PSVtT)) are equal among the vanous peups after adlustments are '
- ! =ade for differences in the values of the independent vanables among ;he p ups, j The matn assumption of the =ethod is that the repession curves for each poup >
- are para!!el (i.e. for each poup ti) the cceincients & are the same. the ccerne:ents
- c. are the sa ne, etc., for a given period). Ancther assumptier is that the rendus:s ;
l of the repessions are normany distributed. This ss mption was venned fer data j !
j in each rone at several periods by plotting the residuals on normal prcbabihty paper i 4 I and using the w elbestablished Kolmogorov Smirnov test. The fact that the residua's .
i were found to be ner=a!!y distnbuted is consistent with the dedinn of others ie.;,
Donovan,1973; McGuire.1974: Campbe!!.1951) who have stu&ed the staustical j
a distnbuttons of 1.'ruted States pound =otion data. The anal assu=pnen is that l
l 4
12 C. B. CROUSE. YOGESH K. VYAS. AND BRUCE A. $CHEL!.
the population variances about the repession lines are equal. Although this as.
sumption was not formally tested, an examinstion of the sample variances in6cated the assumption was reasonable.
The analysis of covariance was applied to the three focal mechanism poups of Northern Honshu data. The equality of the coefncients, b . e,, and d. (where the subscript i = 1,2,3 denotes the couph as well as the equality of the adjusted poup means ofIn[PSV(T)L, were tested at the 95 per cent cenndence level by using the appropriate F ratio (Dixon and .Niassey,1953; Dixon,1955). The results in6cated that, at this conddence 'evel, the adjusted poup means can be considered equal at all psriods. The test for the equality of coetScients fails at enly one peried (T = 1 sech but this isolated case is inconsequential.
The equality of the adjusted coup means is censistent with the 6stributien cf residuals and their sserages for each of the three peups. At each perisi the pre 6eted PSV were compared to the cbserved PSV through the equaticn fcr the normal devaste
, LniPS A - IniPS M, . g 3,n where e,is the ner=a! deviate for the ith datum at a g ven period. (PS A is the etserved PSV at a r.ven peried;(PSD,,is the corresponding preseted PSV vhich i was obtained by substituting the values ef M, R and 6 for the ith datum into ecuatien m: and r. is the per:od. dependent standard errer. The coe:Ceients. a. b, c, and 4. and the standard errors were taken from Tab'e 3. A comparison of the :.
and their asersges. 3. is shewn in Figure 4. The 2. distnbutions cf each focal.
mechanism poup are si=tlar, aniin general. the Ifor each peup are close to :ero, which supper:s the resu!:s of the analysts of eevanance. Thus, it appears that 6fferences in the PSV d.n= from Northern Honshu cant.ot be attr:buted te
'* i
- g . t 8 8 e; m. i i.
- 1. . h. .
4
! s. :: 1: t.
, a.
. s. . p.:- .; .:
- !!- .:!i- F l- 4- i.- ii- L
- l1-i
. .t, l :i t
+
.; :t .
s
' te '
, i.
, t.:
V.' : .
I i
i 1 -
. 2
)
1 4 .e I I *. '- 0 9, 9 4L &1 1 IIC 4 D.str.h;t.;m cf 4 a*4 i as a f.sct ;n c f ;erc.j ar.d ::<sb =tc sr_s= g g {:t t e N:gr e t.
Heasma i.e.:n:a ::ne 5..:: <. ::r fxa..=es e aras:: ::q. ar* r.* *n .a F r.re 1 i
{
I I
( l 1
OROUND MOTIONS FROM SUBDUCTION ZONE EARTHQUAXE3 13 l r
differences in the type of focal mechanism. With the eteeption of the thrust and normal data from the Nankai rene, the data from the other tones were not ;
considered to be sufficient to test the effect oflocal mechanism t>Te on the PSV.
Nonetheless, although the effect of focal mechanism type was not considered when t r.he analysis of covariance was app!!ed to the PSV data from the other zones, the i
separation of the data into the three focal mechanism poups was preserved in the i figures for each sene similar to Firure 4. Testa for differences in the Nankai thrust !
and normal data are reported in the following subsection in the pararaph where !
comparisons between the Northern Honshu and Nankai data are presented. Because :
these tests were also negative, the interpretations of the results for esch zone assume l that focal mechanism t>Te is not an important factor contnbuting to any differences :
ebserved in the PSV. f Co=trarisons elPSV anont suMu:rton rents. Statistical analyses similar to those l
presented in the preceding subsection o ere used to determine whether the PSV .
from the various subduction tones were similar. Rather than forming seven poups of data tone poup for each rene) and testing them together. which would have
(
produced many rejections of the null hypothesis that the adjusted poup =eans were !
similar, the Northern Honshu poup was compared with each of the other six !
subduction.rone poups individually. The results of the analysis of covariance are i summarized in Table 4. The 3. distnbutions and their averages I;r the PSV data l from each subduction zone were also computed using equations W and W: for each i
( observed PSV the corresponding .\f, R, and h values were substituted into the I
repenien equation develcred from the Northern Honshu data (equation tll and l
Table 3) to obtain the predicted PSV datum, which was in turn substituted into equation m to obtain the corresponding a.. In equation @, the period dependent !
standard error, e , was also obtained from Tab:e 3. The ruu!ter e data are i TAst.I 4 Resttts or Anasts et Cossataset l
- t. ****
44 as a+ se es ;# tJ ;( ;# .,
j 1 K a rt. 0 0 o 0 X o 0 0 0 i
- Nanha X X 0 0 X 0 0 0 0 0 !
3 Alaias' O O O O O O X 0 X X l 4 Peru N. Chde : e e 0 0 : a e o !
$ Se 3ntaw 0 0 e o a i X e e i kwa. 1 vce a.1 '
s,ie.,
5turm.Jeaes 0 0 s i e e : a o htt m: s.te' 0 0 i
j X X X X X 0 0 N 4 Meixe 'an 0 a 0 0 X + 3 : a s.t m
$taf W s,tes e a 0 0 0 0 o o o o krt. o.I saes e X 0 a i s s s X X Test d su hywt.tes4. N. at.atei gwp means ci tat /SL B! f:t Nmters H:ssaa and ue .ta sc6ctxa 3:r.e are et.41. C ef:ence 'ese: = 0 35 a at X = re.wt N. a ct 0
- si:en N. l.:eer case i
'ever smufes that the maaty :f ete&uts s. c. ana a :etween Nenhern Herna and t3e .ta sentan ine cunes 4 eecepted at t3e is per cut c:ec:nce tem
' tr4de:ent hta to test eqa.t> ct b . ses 2
l l
l 14 C. B. CROU5E. YOCESH K. YYA3. AND SRUCE A. SCHE!.L summarized in Firates 5 through 10 for the other six subduction renes. Each Crate contains two plots: the upper plot shows the distribution of o. for the thrat. nor=al.
! and stnke4 lip dat4 as a function of period, and the lower plot shows the average l
4.6) for each focal mechanism troup as well as the Ifor a!! of the data combined, i For the .%w Britain /Bougainville and .N!exico subduction tones, the data have also I been separated into two soil classifiestions: he shallow stiff 4 oil sites and the soft.
l soil sites. The ifor each of these soll classes is shown en the lower part of Firates l 9 and 10 for these two renes. Results of the analysis of covariance for these cases I are also presented in Table 4.
. .s
. e !. ,.
y.
s, l'i h le i. I' !.. L }; P_
- u. l.; i' , ;" ,ij "! i:
- f. 1
.. : ; = s .
,+ .
- w. ? . . .;
3 I y 3 2 3 , ,
? -
ra 3 c,,i m m , a . .. m ,, x - . ~ m,.
e,x n u..,,,,,,a u ,....a n m 3,- m. n. m .o.,... . n . m ..,r..e.a
.r. .,. . m , 2...
c.,.s.....
,..,u,,,,.:.,,svu o..c ..n u .n l s. . . I: .. 1.1 .tv h.: . * .
l t l..aj l g*, tj{ 2
- s. l. .
- ?t .i I a -
- . } Q l'r_ F.a._ -
n .a l. .
- 1
.j g ,It 4
. P, !:. j '
i' i,.
b hp F p, b, i .. l.' ; I
'. 6 i* '
l i .
s l : 2 l
- 1. _ -
. Awe. _ ,t i .'
( .
. . , g, . , j . 4. si ..
Fic & Nusai s :ca:e.t:me ta:t see F r rt 5 :*('-3 t:r ** nat : S 1
l l _.
e GROUND MOTIONS FROM SUBDUCTION. ZONE EARTHQUAKES
. - 15 su s. . _
3 e -
- t 6
. E
- s :
. a I
.s
- f
- t-
- 3* i i' h '.
3
.. Il e
.; s :s
. 1 3 , .
2 . *
.n .
3 e
s:- ,
3 :
i i' .
- 8. 1I 3' LG 11, 1 3 33 I3 8I Fic
- Alaskan sutduction tone data. See Firare 5 legend fer espirnatien.
.i.., :.
t.
2 .
- S
- +
- g. ; *t~'
. # : .i t - '
J. J' 1; -.
' .. 4 ' .l
';}
. $ -} . ti ij 8'
- . -i
- j $ s'
- t. :80 e
4
-4 -t I. .
x,
. , l_ *.
. . ~r e-
., i si u n. . as
,g t; 3,g i es is o 1 the La Mohris data. See Ficare $ iegens De further explanationFic. 4. PerwN .
i The resuits indicate that, in general. there are no signiticsnt dif'erences in the data from the Northern Honshu. Kuril < Figure 51 and Sankai iFigure O subduc zones. For the Kuril versus Northern Honshu dats. tne independent variable coetficients ab;. e,. and d,) can be considered equal at the 95 per cent con:1denc level at 7 of 10 periods, and the adjusted coup means of In(PSViT)) ran be censidered equal at all periods except T = 0.5 and 1.0 see iTable 4). At these two periods, the adjusted means for the Kuril zone are pester than those for Northern
- Honshu, which is apparent in firare 5. At the thr.e periods for which the inde.
pendent. variable coefficients were not equal ( T = 0.4,1.0, and . 5 see). picts of th 1
i
(
16 C. B. CROUSE, YOGESH K. VYAS, AND BRUCE A. SCHELL se ws. , s 3 GMaavua . .
3 s 3 e .
8 *
. s s ) . s s ,
s ; e s 3
1-
'* :* -s s j
- 3, i l' -!. s e
~
s : ! 4" !.
... !*._ g J
- 4.'.
. 3'
. , . :; 5 f "
-f t-
-g , ,
Jr J." e s
-s -8 3
1*
i 9 .
... s .
.e I
- W N
, : s' ~
N ..
P j ,. - t., :N ./- .
. .s. .
3' Le LJ Le Le 3, , eB 4B 13 e4 Fic. 9. New Bntains Bougainvite ubduction sone data. The data fr:co the sort soil site iYonki) are identtSed *ita henzontal bars n the ::p f:nre: the casnec une ;n tne bottoa /:(are represents the aserage of these data. The scud ;ine represents the aserage of the sti.ff seti data. 5ymems ::r foes;.
=echanata t>Te are c.ven in Ficare ;
i ws' C' .,s. -*
3
-:r 4.*
- a. : .s 8. * "8 " ."$ *
- j .i -2
.p* de N -1 d I. .
- , a .I' 1 i 4 4. :.
j:. y4 * '
y ;;!: .s *!
-8 9
. , . .: .- y y s .
.o. I s s ,
4 . .
' ,.y- a- -
s 1
.g ,,...(s=...,-e*,.*" 3 1
1
-- - ~
7...; ....-
6 6i &a ae Le e t a #8 68 A8 Flo.10 Metican subduenon tene data. symbol explanat ens are simi:ar to those in Firare 9 esce;t that the soft sod data (desi6T.ated atta bars step /:tarei ans dasned hee @ tim /ts:.rei} are frets =ote than one site.
o, versus each independent variable and the F values from the analysis of covariance were examined. For T = 0.4 and 1.5 see, the plots of o, versus distance. R. generally showed the o, were pester than :ero for R < 200 km and generally less than :ero g for R > 000 km. At T = 1.0 see, the o, show no stri:ing trends with any of the independent variables, the o, are mostly pester than :ero as previously acted.
For the Sankai versus Northern Honshu data, the independent variable coem.
f f
l
(. CROUND MOTIONS FROM SUBDUCTION ZONE EARTHQUAKES 17 cients can be considered equal at all periods. The adjusted group means can be' considered equal except at very short periods (T = 0.1 and 0.2 see) and again at T
= 0.8 sec. At the short periods, the Northern Honshu adjusted poup means are pester; at T = 0.8 see, the Nankai adjusted poup mean is gester (Figure 6).
The data in Figure 6 also suggest some differences in ground motion due to focal-
- mechanism typa. However, when s-talysis of covariance was performed on the L thrust and normal data, the two Isrw poups of data within the Nankai data ba.e . !
the equality of adjusted poup means could not be rejected at the 95 per cent j confidence level for all but two periods (T = 2 and 4 sec).
The data from each of the other nonJapanese subduction zones are generally different from the Japanese data, although undoubtedly sorne of these differences .'
l can be traced to differences in the local geology between the typical Japanese site i
and some sites from other zones. In other instances, the differences appear to be
! !' more related to intrinsic differences in the earthquake source characteristics.
- 1 j The Alaskan data, according to the results in Table 4, are not significantly different from the Northern Honshu data for T
- i 1.0 see. At longer periods (T = 1 1.5,3, and 4 sec), the differences are significant at the 95 per cent confidence level.
At these periods, the adjusted means for the Alaskan data are greater than those for the Nonhern Honshu data. It should be noted that the lack of Alaskan data did i not permit a test of the equality of the coefficients b,, e,, and d,, However, plots (not shown here) of the normalized residuals, o., for the Alaskan data versus each independent variable did not show any correlation or trends with any of these var: ables. The results in Table 4 are consistent with the residual plot in Figure 7.
The interesting iature of this figure is the padual increase in the c, and 5 with increasing period. This trend can be attributed to differences in local geology .i between the Alaskan and Northern Honshu' sites. The sites in both areas are :
underlain by relatively stiff sediments, but the sediments at the Alaskan sites are much deeper. as noted in the previous data base section. The increase in PSV with period at sites with deeper soil deposits has also been observed in the pound motion
- data from the Western United States by Seed et al. (19%) and Trifunae and Lee t
i H979), for example. Taking this effect into consideration, the Alaskan data would then more closely apee with the Northern Honshu data at longer periods. ,
The Peru / Northern Chile data are not significantly different from the Northern Honshu data except at T =
1 0.1,1.0.1.5, and 2.0 see (Table 4 and Figure 5). At the -
j '
three longer periods, the adjusted means of Peru / Northern mfe are significantly l 3
less. Table 4 reveals that the equality of the coeffielents b,, e,, and d/is not generally .
satisfied at the 95 per cent confidence level, but an examination of the residuals j versus each independent variable does not indicate anv obvious trends for T :i 1.0 l
{
see. For T > 1.0 see, trends are not apparent if the data corresponding to the .lf = '
5.3 event, the smallest eanhquake in the Peru / Northern Chile data set,is ignored.
This event accounts for the two largest residuals, os at these periods. i 1
Unlike the Alaskan data, the differences between the Peru / Northern Chile and l j !
3 Northern Honshu data at the longer periods are not thought to be primarily due to i d any differences in local geology. All sites in the Peru / Northern Chile data set except ;
station 4305 !n Lima (the La Stolina site) are underlain by Cascajo a dense to very dense sandy pavel deposit less than about 20 m in depth (Lastrico and Stenge.
j j 1974; Repetto et al,1950). The La hiolina site is underlain by about 25 m of stiff !
silty clay, sandy silt, and dense sandy travel. Thus, this site and the Cascajo sites l 1 are similar in depth and stiffness to the Northern Honshu sites. Although Repetto et cl. (1960) show that the long. period ground motions a' La h!olina are somewhat
)
J l l
i l 4
J l
l_ .
C. B. CROtlSE, YOGESH K. VYAS, AND BRt.'CE A. SCHELL 16 greater than those recorded at the Cascajo site during the 9 November 1974 earthquake and that this difference may be due to the local geologic differences at '
both sites, the residual data in Figure 8 do not indicate that the local geology is the primary factor for the 5 < 0 trend at the longer periods. The two La h!olina data '
points, identified by the horizontal arrows in the top half of Figure 8. follow the average trend dictated by the Cascajo data shown in the bottom half of the figure.
The trends at long periods observed in the Peru / Northern Chile data are also '
seen to some extent in the data from New Britain /Bougainville. Taken collectively, ,
the adjusted means of the New Britain /Bougainville data are similar to those from the Northern Honshu data except at T = 1.0,1.5, and 2.0 see, at which periods the ,
i adjusted means of the New Britain /Bougainville data are signincantlyless. Because :
- 6 of the 16 componerts in this data set were recorded at Yonki, a knuwn soft soil t site, the data were separated into stiff soil and soft soil categories and the analysis +
1 was repeated. The results (Table 4 and Figure 9) show that the stiff soil data are l i
greater than the Northern Honshu data at short periods (T < 1.0 see), while the ;
opposite trend is observed at longer periods. This trend at longer n periods is simij to that noted in the Peru / Northern Chile data. The fact that the coefficienta b,, e i and d did not pass the equality test at T k 0.4 see at the 95 per cent confidence I level does not affect these observations within the ranges of the independent f variables for the stiff soil data.This was verided by examining plots of the residuals versus each of these variables. r
' Another interesting observation in Figure 9 is the similarity in the 5 for the stiff- ;
soil and thrust data and for the soft soil and normal data. These apparent correls- '
(' tions are more of a coincidence rather than redecting possible biases in the soft-
! soil and stiff soil data due to focal mechanism type or vice versa. Twice as many ,
i thrust components (4) were recorded at Yonki, the soft soil site, but the o, for these i data are similar to the e, for the two normal components at this site, as shown in 4 the top half of Figure 9. This comparison further emphasizes the differences in the Yonki data due to local geology. The remaining stiff soil data have approximately the same ratio of thrust to normal components (8:2) as the Northern Honshu zone 193:22) to which it is compared. Thus, the distinct differences between the stiff soil .
New Britain /Bougainville and Northern Honshu data are not the result of any u
I differences in the relative numbers of thrust and normal events between thel The results of the Onal comparison with the hiexico data are presented in Table !
4 and Figure 10. Because of the large disparity in the local geology between the ;
i stiff. soil and soft soil catecories, the results from the combined data base are not j i
too informative. The adjusted group means of the stiff soil data are similar to the :
Northern Honshu data at all periods: however, some caution must be exercised in [
j n interpreting this result because the test for the equality of coefficients b,, e and d,
) failed at the short (T = 0.1 and 0.2 see) and long iT k 1.5 see) periods. At each of !
! these periods, the inequality in the e, coefficient was the reason that the hypothesisl j
regarding the equality of the three sets of coefficients was rejected. A linear trel j
was observed between the residuals and in(R). At the two short periods,the residuats ,
decreased with increasing In[RJ, and at the long periods the opposite trend was i J
observed. Of particular interest are the data recorded in or near >!esico City at
, Tacubaya and Sismex Puebla during the 9 September 1985. earthquake. These da were recorded at the longer epicentral distances within the stiff soil group. The residuals, o,, of these data are between -3 and 0 at the two short periods and
-l < between 0 and +2 at T k 1.5 sec. At these longer periods, these residuals are much
( less than those for the soft soil sites, as illustrated in Figure 10. Although the j
l 1
1
caouso MottoNs raoM steoccTroN zone EARTHQUAKES 19
( equality of the coefficients b , e,, and d, for the soft soil data was rejected at most periods, the reasons were n,ot apparent when the residual plots were examined.
Clearly, most of the residuals were substantially greater than zero at the longer periods, regardless of the values of the independent variables.
I
' The local and regional geologie characteristics,i.e., soft soils and large sedimen-tary basins, are responsible for the large PSV of these Stexico acceleropams.
Approximately half of the hiexico accelerogram data listed in Table 2 were recorded in Alexico City, which is situated on the bed of a Pleistocene lake that occupied the basin of the Valley of Alexico (Tsai,1969). The basin is composed of soft alluvium andlake sediments, which amplified the ground motions at the longer periods (Tsai, 1969; Anderson et al.,1956). The Stinatitlan and Pajaritos acceleropams were recorded near the eastern coast of Alexico adjacent to the Bay of Campeche. This back are region is pan of a large alluvial basin and this basin is probably responsible for the large,long period motions observed in the PSV of these two accelerogams, which were recorded near the middle portion of this basin. By contrast, the accelerograms from the other subduction zones were generally recorded in the fore-are regions at the edges of sedimentary basins, where the sedimentary layers are much thinner than those in the middle of the basins. Long period motions at the edges of the basins are usually not as pronounced as those in the middle, as demonstrated during the 1971 San Fernando. California, eanhquake (Hanks,1975:
Liu and Heaton 1954). During this event, sites at the edges of the Los Angeles and San Fernando Valley basins did not experience large. long period motions, whereas sites near the middle of these basins did experience them.
The results of the analyses of the PSV data were compared to the characteristics of the subduction zones to gain some further physicalinsights for the differences
{ observed in the pound rnotions among the seven subduction zones. Plots of nine of the parameters in Table 1 tage convergence rate, dip, contact width, maximun4 subduction depth, maximum historical eanhquake-3f., maximum rupture length, stress drop. and seismic slip) appear in Figure 11. The two groups of the subduction zones from left to right along the horizontal axis of each plot are in order of decreasing strength of ground motion at stiff soil sites for periods peater than about 0.5 sec. The braces benetth the zone abbreviations tump those zones together in which the strength of motion at these longer periods are similar. This pouping j was based on information presented in Figures 5 through 10 and in Table 4. '
Correlations are not readily apparent between most of the variables plotted in Figure 5 and the observed long period pound motions. If the limited Stexican stiff-soil data are ignored, weak correlations are seen in the stress drop and Jf. p ots.
The maximum stress drops and Af, are somewhat higher for the Alaska Nankai-Kuril Northern Honshu peup than for the Peru / Northern Chile New Britain / i Bougainville group. Although some physical interpretation could be advanced to explain the stress drop observation, the stress drop variations are large, and the i values reported in the literature are probably not consistently determined by one procedure with the same type of data 'o warrant any such interpretation. )
The most anomalous subduction zone in terms of the characteristics listed n i Table 1 is New Britain /Bougainville, the zone with the smaller long period peund motions. This small subduction zone involves a complex interaction of four litho-spheric plates and, consequently,is more complex tectonically than any of the other ;
l tones. Compared to the other six zones. the New Britain /Bougainville zone has the steepest Benioff Wadati zone below the plate interface and has the smallest contact width. Although there are many anomalous characteristics of the New Britain' k
I
( 20 C. B. CROUSE. YOGESH K. VYAS. AND BRUCE A. SCHELL
,: ,="
. . v.
- g. 6 i l a, I ,. .t 3
- 34 '}'
j !: i :18
] "I i
- li 1
- 8 It'!
l.a 1:
l 1.3. . . . - . . 3 .: . . . . . . .
- i - i ! <
- !. s i :
., ...- t .: -
1 - , .
I . :
.. i :4 i .:. . . .
} .!.. . . . - ..,
a' i...
i, I ..
q c: ,
,t.
l i ! ,
1'h" *
' ! ""...l l
,' ( "U 1 -
'. I . I ,.' :
Fic.11. Subductien.sene parameters frem Tab:e 1. The pouping of subduction zones from .' eft to vt a!cng the honzont4 at:5 ts in order of decreasing strenpn of pound motion for penods of 0 3 see and greater. Braces benesta zone abbreviations lump those sones in an:en the strench of motion s .
si=dar. M = Mexico. A = A!aska: N = Nankai: K = Kurd: NH = Northem Honst.u: PC = Pero Scrinen. Chde. NB = New Bnt.un,3 cura:nvnle.
Bougainville subduction zone, the potential physical link between them and the ground motions is not clear.
Discussion The general lack of correlation between the PSV data and the seismotectonic characteristics of the subduction zones, as indicated in Firare 11. may not be surprising. Variables such as convergence rate and age of the subduction zone, although they may correlate with the maximum marnitudes of eanhquakes that have occurred in the subduction zones (Ruff and Kanamori 1950; Heston and Kanamori.1934). may have little or no bearing on the ground motions generr.ted by an earthquake of a givi n moment magnitude. Detailed information, which would be relevant, such as local and average stress drops for many eanhquakes, asperity sizes and distributions, and seismic velocity and Q structure in the upper 150 km, is not available for these regions. Some ir' . nation on source complexity. multi.
plicity, and roughness, which may be usetut, was obtained in a parallel study by Hanzell and Heaton (1965a).
Hanzell and Heaton analyzed data from large magnitude i.4. > -7.5) subduction.
zone eanhquakes and found no correlation between the teleseismic source time functions, trench age, and convergence rate. This tinding is analogous and perhaps consistent with our observations concerning the generallack of correlation between
( the ground motions and the subduction. zone parameters listed in Table 1. two of
GROUND MOTIONS FROM SUBDUCTION ZONE EARTHQUAKES 21 which are the age of the subducted plate and convergence rate. However, the source-( time functions derived by Hanzell and Heaton exhibited similar characteristics for different earthquakes in the same subduction zone. These characteristics tended to vary for different subduction zones. Hartzell and Heaton grouped the zones accord-ing to source multiplicity and source roughness as follows
- 1. South Chile, A!cska
- 2. Aleutians, Kamchatka, Kuril. Colombia Nankai
- 3. N. Honshu, Japan, Tonga Kermadec, Central America
- 4. Central Chile, Peru, Solomon Islands (Neu: Britain /Bougainvil'e ), New Hebrides.
The pouping is in order of decreasing multiplicity and roughness, and the italicized zones are those considered in our study. Central America has not been italicized because our h!exico zone is northwest of the area studied by Hartzell and Heaton.
A comparison of Figures 5 through 10 with the above poupings indicates that there is little correlation between the short period pound motions less than about 0.5 see and the zone grouping. However, at intermediate periods around I sec, some correlation is apparent. For example, the ground motions from Peru / Northern Chile ICentral Chile, Peru. of Hartzell and Heaton) and New Britain /Bougainville (Scl-omon Islands of Hartzell and Heaton) are smaller on the average than those from Northern Honshu. Likewise, the Sorthern Honshu pound motions are smaller on the average than those from Nankai and Kuril, although the differences between the Sankai and the Northern Honshu motions were only statistically signi:icant at a period of 0.3 see. This apparent correlation may be a coincicence because the correlation breaks down at periods around 3 and 4 sec. At these periods, one would expect the correlation to bejust as strong or possibly stronger because the teleseismic data contain some information at these periods ithe period band of the teleseismic data is 2.5 to 50 see). A stronger correlation at these periods would hase offered more persuasive evidence that the source trather than travel path) characteristics at intermediate periods are potentially different in some zones and are thus contributing significantly to the differences in the observed ground motions.
The effect oilocal and regional geology was found to be an important factor also despite the initial attempts to select as much accelerogratn data from stations aether i than those from h!exico) with similar local geology. Geology obviously affected the
$1exico ground motion data, and it probably affected the Alaska data and the I
pound motions recorded at Yonki in New Britian/Bougainville.
The evidence suggesting that local geology affected the New Britain /Bougainville acceleropams recorded at Yonkiis based on a comparison between these data and those recorded at stiff. soil sites within the same zone (Figure 9). Denhatn et al.
(1973) also observed a characteristic shape of the response spectra from the Yonki l
acceleropams. The spectra exhibit a peak centered at 0.2 sec, and Denham et al.
(1973) suggest that the local geology may be the primary factor responsible for this peak. However. the PSV levels observed in the Yonki spectra at the longer perieds i
are much smaller than the PSV levels generally associated with other sotter and er deeper soil sites le.g., Alaska and Alexicot Furthermore. .t is interesting to note the i
sinusoidal character of the average residuals. I. in Figure 9 for both the Yonki and '
stiff soil data. This evidence indicates that other factors, such as the source and or travel path characteristics, are contributing to the spectral characteristics. The anomalous tectonic environment of the New Britain /Bougsinville zone. s dheusied in the preceding section. may be induencing these factors in some systematic manner,
a o
( 22 C. B. CROUSE, YOGESH K. VYAS. AND BRUCE A. SCHELL ACKNOWLEDGh!ENTS The authors wish to thank Slike Leue. George Liang. Doug Coats. Nasur 5foeen.Vann, Yau for providing valuable assistance in developing the cornpu'rer codes and perfor=ing the ne calculations dunns the course of this study. Discussions with Drs. Tom Heaton. Steve Ha Lorden. Nancy 5f ann, and Hiroo Kanarcen were quite helpful. Tne accelerogram data un Japan Port and Harbour Research Insutute. Ann Sion of Kisojiban Consu;tants. t Profen Katayama of the Univnsity of Tokyo, John Anduson of the Univmity of California at San Proteuor Klaus Jacob of Lamont.Doheny Geological Obsuvascry were stestly appreciated.
by Ken Campbeillead to significant improuments in the manuuript.The study was fun Production Research Company and The Eanh Technology Corporation.
RErERtNCEs Andmon. J. G., J. N. Brune. J. Prince. E. Stena. P. Bodin Ed. tors. Armenean Society of Civil Enonens. Nes York. 33-54.
Ando. 51. (19*5). Source mechanisms and tectonic signaticance of historical earthqua Nankai trough. Japan. Tectoncphs.es 2 7, 119-140, ,
Archambeau. C. (19'S). Estimation of non hydrostatic stres identikmq Seumic Gopa ssd Soon.to Bene Gaps. US. Cect Surv. A!uka. Ope't. Pat Rept. M Beaun. J. and K. H. Juob (19M. Proceued strong motion data from subduction zonn:
Lamont.Dohnty Otsuvatory. Columbia Uniursity. Pausades. New York. Cetoter.
Boatwr:cht. J.11950L Preliminary body.wau anabsts of the St. E:ias. Alaska eenhquake of 23.19*9. Ss2 Seus Soc. A'n. 70.419 436.
Bolt. B. A.. C. H. Lob. J. Pennen. Y. B. Tsai. and Y. T. Yeh #195:L 13, Pre'Jminary report 1 strong motion may in Taiwan. Eanhquake Enoncering Research Center. Repon EER Unnersity of California. Berie:ey, Californta.
4 f Brady, A. G. and V. Peru (1977L Stroeg.: notion vanhquake acteWrestams, dipt:tation r
records from Li:na. Peru: 1951 to 19*4. CS. Geet Su . . 0;e+FJe R4; . 774J7 Campbou. K. W. t195D. Nur source attenuation of tseak honzental acce'. nations. Ba2 Se
- 71. 0039-2070 Chael. E. P. and G. S. Stewart (1HD. Recent urge unhquakes akng the 5!!dde A=n:ca tre
- 87. 309-333 their impheationt for the subduction procen.J Geephn Res Chang. L S. 51954L Ano!>su of Aceciemgas -a /or Strong.Afotion Estr4woces m Ts;2an. V A. 3. C. and D. Inst:tute of Eanh scionen. Academia $inica. Repuc.ie of Ch:na. Slay.
Chec. L 5. and H. C. Ch29 (19M. Aso >s s o/ Accelmgasm.e fo Stay.
Couch R., R. Whitaett. B. Huehn. and L Briceno.Guarupe i198D IS4.*03-7:6.
Crouw. C. B J. A. Hikman. B. E. Turner. and G. R, Stanm i19fa Comritat.cn. suus=ent ud uparmon of the strong utthquake stound motion data bue. U.S. Naclear Regulatory Commi NUREGs CR.1560.
Davies. J., ' . Sykes. L Houp, and K. Jacob (1991). Shumarin wism2c gap. Aluan Pentnsu Res 86 great earthquakes, tectonic atting, and mdence for high utsmic potent:al. / Geeps 3821 3333 Dean. B. W. ed C. L Drake t1975L Focal muhanism solutions and tutonics are. /. G
- J Joe. !cr Es?rscae E .g 4.15-50 Denham. D , G. R. 3rnau. and 1. B Euringham 11973L Some strons. motion rou! s from Pa Guinea 1967-1970. Pmc. Air 4 Word Ces/. E.vr%44 Eng 1. 610-413.
j Dewey, J. N. and W. Spence i19*91. Seismic gaps and source tones of ruent large unhquue Peru. Pure Appa Geopsn 117. 1146-1171 i
I l
GROUND MOTIONS FROM SUBDUCTION ZONE EARTHQUAKES 23 Dixon. W. J. (Editor) (1935). B.\fDP Statur col Sorracre. University of California Press.
f Dixon. W. J. and F. J, Massey, Jr. (1953L Introdur:on to Starntical Ana!>ss. 4th ed. McGraw Hill
(' Book Company. New York.
Donovan. N. C. 0973L A statisneal evaluation of strong motion data including the February 9.1971 San Fernando ear:hquake. P oc. Fifth World Conf. Earthquake Eng. 1.1:52-1:61 D :eworuki. A. M. and J. H. Woodhouse 0953L An expen:nent in systematic study of glocal seistnie;ty:
centroid. moment tensor solutions for 201 moderate and large earthquakes of 1961. 4 Geerny s. Res.
- 88. 3247-3271.
- EERI t1965L Earthquake in Chile. March 3.1955, Earrhei.ake Engineenns Resecrch Icar:r;.:e .Vea s.er er 19.
- .
- aph.ss. Res Aleutian aret tectonic blocss, episod;e 8 2. :13-230. .
- 29. 061-;7 .
- n.c. A t
- 13*31. Crat: