,... .

' +

Strong Ground Motion Attenuation

Relationships for Subduction Zone
Earthquakes
R.R. Young·s
8.-J. Chiou
Geomatrix Consultants

W.J. Silva
Pacific Engineering and Analysis

J.R. Humphrey
Lahontan GeoScience Inc.

ABSTRACT quakes are distinguished from shallow crustal earthquakes

that occur in the upper 20 to 25 km of continental crust
Yx/e present attenuation relationships for peak ground aced- (such as the 1989 M 7.1 Lorna Prieta, 1992 M 7.3 Landers,
' ;·arion and response spectral acceleration for subduction and 1994 M 6.7 Northridge earthquakes in California).
:one interface and intraslab earthquakes of moment magni- Published attenuation relationships for subduction zone
tude M 5 and greater and for distances of 10 to 500 km. The earthquake motions (e.g., Iwasaki et al., 1978; Sadigh, 1979;
:elationships were developed by regression analysis using a Vyas etal., 1984; Krinitzkyetal., 1987; Crouse eta!., 1988;
random effects regression model that addresses criticism of Youngs eta!., 1988; Crouse, 1991) typically indicate that at
earlier regression analyses of subduction zone earthquake distances greater than 50 km from the earthquake rupture,
motions. We find that the rate of attenuation of peak ground motions from these earthquakes are substantially I
motions from subduction zone earthquakes is lower than larger than those from shallow crustal earthquakes in active
that for shallow crustal earthquakes in active tectonic areas. tectonic regions. However, some investigators believe that
!"his difference is significant primarily for very large earth- the ground motions from interface earthquakes and shallow ·
quakes. The peak motions increase with earthquake depth crustal earthquakes are similar, at least in Japan (Fukushima j
and intraslab earthquakes 'produce peak motions that are and Tanaka, 1990; Iai eta!., 1993). Both Fukushima and I
about 50 percent-larger than interface earthquakes. Tanaka (1990) and Iai et al (1993) concluded that the rea-
son previous researchers found lower attenuation rates for
INTRODUCTION ground motions from Japanese earthquakes was that theu
employed a single stage regression analysis which was biased '1' I
1

This paper presents attenuation relationships for subduction by correlations in the data. When Fukushima and Tanak
zone earthquakes. Two types of subduction zone earthquakes (1990) and lai et a!. (1993) used a two-staged regression I
are considered, interface earthquakes and intraslab earth- technique similar to that employed by Joyner and Boore
quakes. Subduction zone interface earthquakes are shallow (1981), they found attenuation rates for shallow (depth less I
angle thrust events that occur at the interface between the than 100 km) earthquake ground motions to be similar to
subducting and overriding plates. Examples include the
1964 M 9.2 Alaskan earthquake, and the 1985 M 8.0 Val-
that reported for crustal earthquakes in the western United
States. Therefore, Fukushima and Tanaka (1990) and Iai et
I
pariso, Chile, and Michoac~, Mexico, earthquakes. Sub- al (1993) combined shallow crustal and subduction zone I
duction zone intraslab earthquakes occur within the earthquake strong motions into a single data set to develop
subducting oceanic plate and are typically high-angle, their attenuation relationships. I
normal-faulting events responding to downdip tension in In this paper we restrict the data to subduction zone
the subducting plate. Examples include the 1949 mb 7.1 and
1965 M 5 6.5 earthquakes in the Puget Sound region of
interface and intraslab strong motion recordings. To address
the concerns raised by Fukushima and Tanaka (1990) and Iai
I
Washington State. In this paper, these two types of earth- (1993) we employ the random effects regression model of I
58 Seismological Research Letters Volume 68, Number 1 January/February 1997
. "' if ·- :,~/I' Ct- ~. I
 I .

Abrah~son and Youngs (199 2), whi ch is equi

valent to a is less than 20 m and a significant velo
tWO-staged regression analysis (Brilling city cont rast may exist
er and Preisler, 1985; with in 30 m of the surface.
Joyner and Boore, 1993). In addi tion
, we use numerical sim- The data used in the analysis were restr
ulations of large interface eart hqu akes icted to free-field
to aid in judg ing the recordings from mag nitu de 5.0 and
appropriateness of the atte nuat ion mod greater events. Free-field
els. The relationships recordings are considered to be reco
given in this pape r were first pres rdings obta ined at the
ented by Youngs et al. base men t or the first floor levels of
(1993). buildings less than four
stories in height. In addi tion , data were
excluded if the qual-
STRONG MOTION DATA BASE ity of the recorded acceleration time
histo ry was poo r or if a
port ion of the mai n shak ing was not
recorded.
'1 The data set analyzed in this stud y is Figure 1 shows the mag nitu de-d istan
sum mar ized in Table 1. ce scattergram of
_ ~~ · The mag nitu de measure used was the stro ng mot ion data set collected
Han ks and Kanamori's for analysis. The largest
J' Jt (1979) mom ent mag nitu de .M. Source parameters grou p of data is soil site data for inte
rface events, primarily
earthquakes (epi cent er loca tion , foca of the from Japan. A large port ion of the
l dep th, mag nitu de, and rock data are from three
focal mec hani sm) were com pile d Mexican subd ucti on zone eart hqu akes
from pub lishe d special recorded at the Gue r-
studies or the Har vard cent roid mom rero Array. The 199 2 M 7.0 Petrolia,
ent tensor solutions. Cali forn ia eart hqu ake
The Har vard solu tion s were used only mai nsho ck was also incl uded as an inter
for seismic mom ent face event.
M 0 and focal mec hani sm. If no special
stud y was foun d for ANALYSIS OF PEAK HORIZONTAL
an event, then eart hqu ake location
and mag nitu de given in ACC
the Inte rnat iona l Seismological Cen ELERATIONS
ter or Nati onal Eart h-
quake Info rma tion Cen ter catalogs
were used. If seismic Atte nuat ion relationships for hori zont
mom ent was not repo rted , then the al peak grou nd accel-
surface wave mag nitu de erati on (PGA) from subd ucti on zone eart
f'is was used, assuming that it is equivalent to mom hqu akes were eval-
ent mag- uated by perf orm ing regression anal
nitu de in the mag nitu de range of 6 yses on the empirical
to 7.5 (Ha nks and Kan- data. Because there are insufficient data
amori, 1979). If only bod y wave mag to rend er stable esti-
nitu de mb was reported, mate s of the regression parameters indi
then mb values in the range of 5 to vidually for each of~
6 were conv erte d to M the six data groups, except for the inte
using the relationship M = 1.8m b- 5 rfac e soil data, we per-
5 4.3 prop osed by Wyss formed a join t regression analysis. The
\ and Hab erm ann (1982) and the resu cons tant s dete rmin ed
lti-ng value taken to be from the analyses incl ude the atte nuat
equivalent to mom ent mag nitu de. ion relationship coeffi- j(
The differentiation cien ts for the selected reference grou p and
between interface and intraslab even parameters repre-
ts was don e on the basis sent ing the pert urba tion s of the othe
of the faulting mec hani sm, whe n repo r five grou ps from the
rted, or on the basis of reference atte nuat ion relationship.
the focal dept h, with events below a A set of indi cato r vari-
dep th of 50 km consid- able s were used w identifY data from each
ered to be intraslab events. Tichelaa group: Z, indicates
r and Ruf f ( 1993) indi- source type (0 for interface events and
cate that interface eart hqu akes wor ldw 1 for intraslab events),
dept hs shallower than 50 km.
ide nearly all occu r at zds indicates deep soil cond ition s (I for
deep soil sites and 0
We characterized source-to-site dista othe rwise), Zss indicates shallow stiff (I
nce in terms of the for shallow soil sites
closest distance to the rupt ure surface, and 0 otherwise), and Z, indicates rock
rrur If the rupt ure sur- sites (I for rock sites
1._ face was not defi ned for an and 0 otherwise). The basic regressio
event, then hyp ocen tral distance n mod el follows the
"' was used as the source-to-site dista form used previously by Youngs et al. (198
nce. This is not expected 8} and Crouse
to intro duce a significant bias because (1991}:
rupt ure surfaces were
defined for nearly all of the large even

l
ts, and for the small
events the difference between the min
ture and hyp ocen tral distance is sma
source-to-site distances for the reco
imu m distance to rup-
ll in com pari son to the
rdings.
Based on pub lish ed info rma tion on
site cond ition s, the
ln(PGA) ij~~~ c; +'"'),
+C, M, + ~;
+ 'c;- M,

recordings were classified into thre ( +C5 Zr+ Cc/ !i+C loZs s +1], +EiJ,
e groups: rock, shallow
stiff soil, and deep soil sites. Roc '-------·- --- ,_
expected to be simi lar to typical rock
k site cond ition s are
cond ition s in the Cali-
c; = C1 + C6Z,
0)
fornia stro ng mot ion database, cons C~ =C3 +C7 Z,
isting of at mos t a few
feet of soil over wea ther ed rock. This
ered to be cons isten t with Boo re et
classification is consid-
al (199 3) Site Class A
c; =C4 +C8 Z,
near the bou nda ry with Site Class B.
Dee p soil sites are those
where the dep th to bedr ock is expected where i is the eart hqu ake index, j
to be greater than 20 is the reco rdin g stati on
m. This site das5ification is consider inde x for the ith event, PGA (in unit
ed to be cons isten t with s of g) is the geometrical
Site Cla~s C pres ente d by Boore et mean of the two hori zont al com pon
al. (199 3). The shallow ents of peak grou nd
stiff soil classification represents sites acceleration, M is mom ent mag nitu
where the dep th of soil de, rrup is the source-to-
site distance (in kilometers), His foca
l dep th (in kilometers),

Seismological Research Letters Volume

I 68, Number 1 January/February 1997
59
 TABLE 1
List of Earthquakes Used to Develop Attenuation Relationships
H Distance
Earthquake Date Lat. Long. (km) FT1 M Range Number of Records2
RK ss OS
Alaska
Alaska 1964.06.05 60.4 -146 16 t* 5.2 27.6-27.6 1 0 0
Alaska 1964.06.05 58.1 -152 13 t* 5 49.Q-49.0 1 0 0
Alaska 1965.09.04 58.3 ·-153 32 t 6.8 68.3--£8.3 1 0 0
Adak 1971.05.02 51.4 ·-·177 38 t* 6.8 77.3-77.3 1 0 0
Shumagin Is 1974.04.06 54.9 ···160 37 t 5.6 65.5--£5.5 1 0 0
Shumagin Is 1974.04.06 54.9 ·-160 40 t 5.8 64.7--£4.7 1 0 0
Alaska 1975.05.18 63.2 ·-150 106 n 5.4 144.1-144.1 0 0 1
St Elias 1979.02.28 60.6 -·142 13 t 7.5 43.Q-101.0 1 0 2
Alaska 1983.02.14 54.7 -159 25 t 6.5 40.4-126.0 2 1 0
Alaska 1983.02.14 54.9 --159 25 t 6 37.3-121.5 2 1 0
Alaska 1983.06.28 60.2 -·141 19 t 5.9 23.1-23.1 1 0 0
1985.10.09 54.7 --159 15 t 6.6 19.3-50.1 2 0 0
1985.10.26 54.8 --159 18 t 5.3 20.5-20.5 1 0 0
1985.11.14 54.7 --160 19 t 6 24.5-24.5 1 0 0
Andreanof Is 1986.05.07 51.5 ·-·175 33 t 7.9 33.Q-33.0 0 0 1
1987.06.21 54.1 --162 33 t 6.5 47.0-47.0 1 0 0
Chile
central Chile 1945.09.13 -34 -71 100 n* 7.1 107.7-107.7 0 1 0
central Chile 1952.04.29 -35 -72 10 t* 6 169.6-169.6 0 1 0
central Chile 1953.09.04 -33 -72 50 n* 6.4 144.8-144.8 0 1 0
central Chile 1958.09.04 -34 -70 15 t* 6.8 81.8-81.8 0 1 0
La Ligua 1965.03.28 -32 -71 72 n 7.4 145.7-145.7 0 1 0
central Chile 1967.09.26 -30 -72 48 n 5.6 396.Q-396.0 0 1 0
Valparaiso 1971.07.09 -33 -71 42 t 7.8 101.0-101.0 0 1 0
central Chile 1973.10.05 -33 -72 23 t* 6.7 130.4-130.4 0 1 0
central Chile 1974.11.12 -33 -71 90 n* 6.2 94.Q-96.7 1 1 0
central Chile 1978.12.21 -36 -72 46 t 5.8 66.3-127.7 0 0 3
central Chile 1979.07.05 -32 -71 52 n* 5.8 67.2-113.3 2 0 1
central Chile 1981.11.07 -32 -71 65 n 6.9 74.6-178.7 1 1 4
Valparaiso 1985.03.03 -33 -72 31 t 7.9 23.3-256.6 7 1 15
Valparaiso AS 1985.04.09 -34 -72 44 t 7.1 66.3-219.2 1 1 4
Cascadia
Puget Sound 1949.04.13 47.2 -123 56 n 7.1 61.9-73.4 0 0 2
Seattle 1965.04.29 47.4 -122 63 n 6.7 88.9-88.9 0 0 1
Petrolia 1992.04.25 40.3 -124 11 t 7 8.5-71.1 2 4 8
Japan
Chiba Pref 1956.02.14 35.7 140 45 n 6 46.6-46.6 0 1 0
Japan 1962.04.23 42.2 144 60 n 7 109.9-109.9 0 1 0
lbaragi 1963.05.08 36.4 141 40 t* 6.1 65.5--£5.5 0 1 0
Chiba 1963.08.04 35.4 140 39 t* 5.1 48.4-48.4 0 1 0
lbaraki 1964.11.14 36.5 141 69 n 5.1 69.0--£9.0 0 1 0

60 Seismological Research Letters Volume 68, Number 1 January/February 1997

 I'
-
if
,,
TABLE 1 (CONTINUED)
List of Earthquakes Used to Develop Attenuation Relationships
H Distance
FT1 M Range Number of Records2
Date Lat. Long. (km)
Earthquake ss OS
RK
72.7-72.7 0 1 0
1967.11.19 36.4 141 48 t* 6
lbaragi 59.Q-127.0 0 1 2
32.3 133 37 t 7.4
Hyuganada 1968.04.01 0 0 1
132 40 t 6.3 111.3-111.3
1968.04.01 32.3 10
Hyuganada AS 8.2 102.9-550.9 0 5
1968.05.16 40.7 144 20 t
Tokachi-Oki 201.7-201.7 0 0 1
40.3 143 20 t 5.1
Tokachi-Oki AS 1968.05.17 0 0 1
143 31 t 7.1 191.9-191.9
1968.06.12 39.4 1
lwate-Oki 6.1 84.8-84.8 0 0
1968.07.01 36 139 68 n*
Saitama 51.7-51.7 0 0 1
33.3 132 48 n 6.8
W. Shikoku 1968.08.06 0 1
39 t 6.3 66.5-66.5 0
1969.04.21 32.2 132
Hyuganada 240.9-240.9 0 1 0
43.6 147 30 t 8.2
South Kurils 1969.08.00 0 1 1
25 t 6.4 43.9-124.9
1970.01.20 42.4 143
Hokkaido 76.4-126.1 0 1 1
39.8 142 75 n 5.8
lwate 1970.04.01 0 1 1
47 t 7 55.Q-71.7
1970.07.25 32.1 132
Hyuganada 72.2-72.2 0 0 1
32.1 132 47 t 5.9
Hyuganada As 1970.07.26 0 0 1
137 44 n 6.1 87.4-87.4
1971.01.04 34.4 1
Aichi 7.1 159.0-263.5 0 1
1971.08.02 41.2 144 45 n
Erimomisaki 42.4-42.4 0 1 0
35.9 141 40 t 5.2
Chiba 1971.10.11 0 1 2
141 50 t 7.4 295.1-307.6
1972.02.29 33.2 2
Hachijojima 6.4 97.2-203.2 0 0
1972.03.19 40.9 142 80 n
88.4-88.4 0 1 0
42.6 145 63 n* 5.8
Kushiro, j044 1972.05.11 0 2
7.5 286.1-417.9 0
1972.12.04 33.2 141 50 t
Bonin 99.0-99.0 0 1 0
43 146 41 t 7.8
Nemuro-Oki 1973.06.17 0 1 0
141 49 t 5.6 65.6-65.6
1974.03.03 35.6 0
Chiba 6 83.6-83.6 0 1
1974.07.08 36.4 141 45 t
lbaragi 126.8-251.0 0 1 1
42.5 142 125 n* 6.5
Tomskomai 1974.11.08 0 1 0
141 44 t 5.6 69.4-69.4
1974.11.15 35.8 10
Chiba 6.5 121.9-537.0 0 2
1978.02.20 38.8 142 50 t
389.3-389.3 0 1 0
44.2 149 31 t 7.5
South Kurils 1978.03.24 0 4 13
40 t 7.6 93.Q-475.1
1978.06.12 38.2 142
Miyagi-Ken-Oki 65.3-196.7 0 1 3
36.6 141 40 t 5.9
1979.07.11 0 0 1
70 n 6.1 162.0-162.0
1979.07.13 33.9 132 3
5.3 89.1-114.1 0 1
1980.09.23 36 140 80 n
161.0-744.2 0 1 6
42.4 142 130 n 6.8
1981.01.23 0 0 5
60 n 6.3 118.0-220.7
1981.12.02 40.9 143 5
6.9 75.5-207.7 0 1
1982.03.21 42.1 143 40 t
71.3-129.8 0 0 2
42.2 143 18 t 5.5
1982.03.21 0 1 5
30 t 7 119.9-356.9
1982.07.23 36.2 142 5
E off Kanto 7.7 81.7-166.6 0 1
1983.05.26 40.4 139 14 t
Nihonkai-Chubu 121.2-233.7 0 1 2
40.2 139 23 t 5.9
W off NTohoku 1983.06.09 0 1 1
14 t 5.9 121.5-166.7
1983.06.09 40.2 139 4
157.4-272.6 0 1
41.3 139 20 t 6.8
W off NTohoku 1983.06.21 0 1 1
43 t 5.5 87.2-186.1
1984.01.17 36.5 141

Seismological Research Letters Volume 68, Number 1 January/February 1997 61

 TABLE 1 (CONTINUED)
List of Earthquakes Used to Develop Attenuation Relationships
H Distance
Earthquake Date Lat. Long. (km) FT1 M Range Number of Records2
RK ss OS
1984.01.17 36.5 141 43 t 5.7 88.2-187.5 0 1 1
1984.03.06 42.5 143 118 n 5.7 125.2-202.6 0 0 2
SE off Kyushu 1984.08.06 32.4 132 33 n 6.9 93.Q-184.1 0 0 2
1984.09.18 34.1 142 13 n 6.8 223.5-232.7 0 1 1
1985.03.28 40.3 141 164 n 6.3 178.6-357.8 0 1 3
1985.05.13 33 1:33 39 n 5.5 114.2-167.4 0 0 2
1986.02.12 36.4 141 44 t 6.2 73.5-160.4 0 1 2
1986.11.28 36.4 141 42 t 5.8 78.8-177.1 0 1 3
1987.01.14 42.5 143 119 n 6.8 126.2-513.4 0 1 6
1987.02.06 36.9 142 30 t 6.4 160.0-174.6 0 1 1
E off Fukushima Pr 1987.02.06 37 142 35 t 6.7 94.4-402.3 0 1 7
1987.03.18 32 132 48 n 6.6 76.7-226.1 0 0 2
1987.04.07 37.3 142 44 t 6.6 103.9-288.7 0 1 6
1987.04.16 37 142 45 t 6 157.0-159.6 0 0 2
1987.04.22 37.1 142 47 t 6.6 80.9-260.1 0 1 5
1987.05.11 38.9 142 50 t 5.8 123.1-206.6 0 0 4
1987.09.24 36.6 141 41 t 5.9 64.1-202.7 0 1 5
Mexico
Mexico 1962.05.11 17.3 -100 40 t 7.2 248.6-248.6 0 0 1
Mexico 1962.05.19 17.1 -100 33 t 7.2 262.3-262.3 0 0 1
Mexico 1962.11.30 17.3 -99 57 n* 5.8 245.1-245.1 0 0 1
Mexico 1964.07.06 18 -101 100 n 7.4 242.0-252.9 1 0 3 I
Mexico 1965.08.23 16.3 -96 16 t 7.4 473.4-481.3 1 0 1
Mexico 1965.12.09 17.3 -100 57 n* 6.3 77.0-261.0 1 0 1
Mexico 1966.04.11 18 -103 30 t* 5.5 100.0-100.0 1 0 0
Mexico 1966.09.25 18.3 -101 79 n 5.7 140.7-140.7 1 0 0
Mexico 1968.07.02 17.6 -100 41 n 6.5 104.8-236.6 1 0 1
Mexico 1968.08.02 16.6 -98 16 t 7.3 192.6-330.0 2 0 1
Mexico 1971.09.05 17.1 -100 50 n* 5 56.7-56.7 0 0 1
Mexico 1973.08.28 18.3 -97 84 n 7.1 156.1-296.6 0 0 4
Mexico 1975.03.14 16.6 -93 155 n 5.6 158.5-158.5 0 0 1
Mexico 1976.06.07 17.4 -101 45 t* 6.4 109.2-109.2 0 0 1
Mexico 1978.03.19 17 -100 36 t 6.6 43.9-43.9 0 0 1
Oaxaca 1978.11.29 16 -97 18 t 7.7 122.3-424.2 1 1 4
Oaxaca 1st AS 1978.11.29 16.2 --97 33 t 5.2 107.5-437.2 1 1 1
Oaxaca 2nd AS 1978.11.29 16.2 --97 22 t 6 102.3-441.7 1 1 1
Guerrero 1979.03.14 17.5 -101 20 t 7.4 71.4-351.1 2 1 11
Mexico 1981.09.17 16.8 --99 22 t* 5.4 22.3-54.6 0 0 2
Playa Azul 1981.10.25 17.8 -102 20 t 7.2 23.6-422.2 3 1 13
Michoacan 1985.09.19 18.2 -103 18 t 8 12.9-358.7 16 3 9
Michoacan AS 1985.09.21 18 -101 15 t 7.5 24.4-191.6 11 1 1
Michoacan AS 1985.10.29 17.6 -103 20 t 5.9 59.1-59.1 1 0 0

62 Seismological Research Letters Volume 68, Number 1 January/February 1997

I
I. TABLE 1 (CONTINUED)
List of Earthquakes Used to Develop Attenuation Relationships
I H Distance
I Earthquake Date Lat. Long. (km) FT1 M Range Number of Recordr
RK ss OS
I Michoacan 1986.04.30 18 -103 20 t 6.9 38.2-90.6 2 0 0
I Michoacan AS
Guerrero
1986.05.05
1986.05.29
17.8
16.9
-103
-99
20
36
t
t
5.9
5.2
39.6-39.6
49.3-95.1
1
5
0
0
0
0
I Peru
1989.04.25 16.6 -99 19 t 6.9 18.4-197.4 14 0 2

Peru 1947.11.01 -11 -75 30 t* 7.7 271.Q-271.0 0 1 0

I Peru
Peru
1951.01.31
1952.08.03
0
0
0
0
50
50
n*
n*
6
5.3
116.0-116.0
125.0-125.0
0
0
1
1
0
0
I Peru 1957.01.24 0 0 50 n* 6.3 120.D-120.0 0 1 0
Peru 1957.02.18 0 0 100 n 6.5 152.0-152.0 0 1 0
I Peru 1966.10.17 -11 -79 24 t 8.1 164.9-164.9 0 1 0
I Peru 1970.05.31 -9.4 -79 56 n 7.9 259.4-259.4 0 1 0
I Peru 1971.11.29 -11 -78 54 n 5.3 131.7-131.7 0 1 0
Peru 1974.01.05 -12 -76
I 98 n 6.6 131.1-131.3 0 2 0
Peru 1974.10.03 -12 -78 27 t 8.1 70.8-73.8 0 2 0
I Peru, AS 1974.11.09 -12 -77 30 t 7 68.3-74.6 0 1 1
Solomon Islands
Long Island 1967.11.14 -5.5 147 194 n* 5.8 243.4-243.4 0 0 1
Long Is 1968.04.29 -5.4 146 31 t* 5.8 101.1-101.1 0 0 1
Long Is 1968.06.03 -5.5 147 182 n* 5.6 226.5-226.5 0 0 1
N.Huon 1968.06.17 -6.3 147 106 n* 5.2 124.0-124.0 0 0 1
New Britain 1968.09.16 -6.1 149 49 n* 6.3 313.5-313.5 0 0 1
Arona 1969.01.07 -6.2 146 111 n* 5.1 122.3-122.3 0 0 1
Umboi Is 1968.03.10 -5.6 147 194 n* 6 232.1-252.6 0 0 2
Umboi Is 1969.06.24 -5.9 147 117 n* 5.2 152.6-153.9 0 0 2
Lae 1969.08.02 -6.5 147 33 t* 5.1 40.1-40.1 0 0 1
Dantu 1969.08.03 -4.3 153 59 n* 5.4 112.9-112.9 0 0 1
Taki 1969.09.07 -6.6 156 174 n* 5.1 179.1-179.1 0 1 0
Ulingan 1970.10.31 -4.9 145 42 t 7 162.D-162.0 0 0 1
Wasu 1971.02.12 -6.3 147 123 n* 5.8 135.9-142.7 0 0 2
Wasu 1971.02.13 -6.1 146 114 n* 5.4 119.7-157.9 0 0 2
Madang 1971.03.13 -5.8 145 114 n* 6.9 142.2-142.2 0 0 1
New Britain Is 1971.07.14 -5.5 154 43 t 8 153.Q-153.0 0 1 0
Annan berg 1971.07.19 -4.9 145 75 n* 5.8 232.1-232.1 0 0 1
New Ireland Is 1971.07.26 -4.9 153 43 t 8.1 251.Q-251.0 0 1 0
New Britain Is 1971.09.14 -6.5 152 22 t* 6.3 258.7-258.7 0 0 1
Lae 1971.09.25 -6.5 147 111 n* 7 119.1-119.1 0 0 1
Kokopo 1971.10.14 -4.4 152 25 t* 5.6 38.3-38.3 0 0 1
Buka Is 1971.10.28 -5.6 154 107 n* 6.5 271.1-271.1 0 0 1
Long Is 1972.11.05 -5.4 147 229 n* 5.4 273.5-273.5 0 0 1
Marien berg 1973.08.13 -4.5 144 109 n* 6.3 304.1-418.6 0 0 2
Saidor 1974.03.25 -6 146 110 n* 5.4 113.1-113.1 0 0 1

Seismological Research Letters Volume 68, Number 1 January/February 1997 63

'·

TABLE 1 (CONTINUED)
List of Earthquakes Used to Develop Attenuation Relationships
H Distance
Earthquake Date Lat. Long. (km) f11 M Range Number of Records2
RK ss OS
Saidor 1974.09.20 -6.2 146 105 n* 6.1 106.D-1 06.1 1 0 1
Solomon Is 1981.12.13 -6.4 155 50 t 5.8 79.6-79.6 1 0 0
Solomon Is 1981.12.13 -6.3 155 48 t 5.4 78.6-78.6 1 0 0
1. t = interface thrust, n = intraslab, * indicates mechanisms inferred by depth
2. RK =rock, SS =shallow soil, OS =deep soil

9 rr~,--~~~~~~,--~~~~ I I

Interface Intra slab

8 - -
0

- -
0

- -

5 f- -
• Rock • Rock
0 Shallow soil 0 Shallow soil
0 Deep soil 0 Deep soil
4 ~~~--~~_.~~~~~~~~~ I I

5 10 20 50 100 200 5001000 5 10 20 50 100 200 5001000

Distance (km) Distance (km)

A. Figure 1. Scattergram of subduction zone PGA data set. ~;~~¢-i\ -l> i r- \" ~¥:P'
!\ . . ___ ,.
1
and Ck, k = 1 to 10, are coefficients determined by regression !motio'ils. The terms 17; and £ij are assumed robe independent
analysis. The terms Hand Z,, are additions to the Youngs et 1 normally distributed variates with variances and if, r
al. (1988) model. Crouse et aL (1988) first proposed that / respectively. Following the model developed by Youngs et al.
peak motions are proportional to the depth of the event. In {1995), variance terms 't' and 0" were assumed to be linear
contrast, Youngs et al. {1988) found that the depth effect ~ functions of magnitude. The regression coefficients and the
observed by Crouse et al. (1988) could be explained by error terms 17; and £ij were obtained by the random effects
accounting for a difference between interface and intraslab regression algorithm described in Abrahamson and Youngs
events. In this analysis we find that both effects are signifi- {1992). Figure 2 compares the fitted relationships to PGA
cant. Recently, Molas and Yamazaki (1995) also report that data from interface earthquakes recorded on rock and deep
soil sites.
{ there is a significant correlation between depth and peak
amplitude for Japanese subduction wne strong motion data. The recorded strong motion data shown on Figure 2
The error term in (1) is partitioned into an inter-event com- display a large difference between rock and soil site PGAs at
ponent 17; representing the earthquake-to-earthquake vari- all distances. The attenuation relationships fit to the data
ability of ground motions, and an intra-event component£;· predict that the ratio of soil to rock PGA increases as the
representing within earthquake variability of ground ground motion level increases. This result is contrary to what

64 Seismological Research Letters Volume 68, Number 1 January/February 1997

I
I
I .5 •
I
,--....
tl) •
.._., .2
• •
~
I .
"+->
0
...,
.1

I d
!-..
.....,
~
.OS
0
~
I (.)
(.)
.02 0

I
~
.01 • •
0

~~ .005 Interface
I n.. M 5.3-5.7
Interface
M 5.8-6.:;>
Interface
M 6.3-6.7
0 Rock (33) 0 Rock {27) 0
.002 Rock (23)
I + Deep Soil (27) + Deep Soil (40) t Deep Soil ( 42)
.001
I
.5 0

,----'
tl)
..__... .2
~ .1
·-'
...... 0
tj
;..._ .05 0
,,
--c:
1...
l.l
.02
0

""!'
.01
...!<
d
c:J .005 - Interface
n.. M 6.8-7.2
- Interface
M 7.3-7.7
lnlerface
M 7.8-8.2
0 Rock (21) 0 Rock (16) 0 Rock (22)
.002
• Deep Soil (26) t Deep Sed (28) t Deep Soil (25)
.001
10 20 so 100 200 500 1 0 20 50 100 200 50010 20 50 100 200 500
Distance (km) Distance (km) Distance (km)
..._Figure 2. Comparison of PGA values predicted using regression rnodcl (1) and the empirical data for interface earthquakes. Numbers
in parentheses give the average depth of earthquakes in each data subset. The solid line is the attenuation relationship fit to deep soil site
data and the dashed line is the relationship fit to the rock site data.

one would expect because nonlinear soil effects should from the 1985 Valpariso and Michoacan earthquakes using
reduce soil amplification as the level of shaking increases. We this simulation model.
examined the expected difference between soil and rock Figure 3 compares the results of the simulations for
PGA by performing numerical simulations of ground rock, shallow soil (20 ft depth) and deep soil (120 ft and 500
motions on rock and soil sites from a M 8 subduction zone ft depths) to the results of the regression analysis using (1).
earthquake using the finite-source form of the stochastic There is good agreement between the simulations and the
ground motion model (Silva and Stark, 1992). The simula- empirical model at distances greater than 50 km. At smaller
tions incorporated site effects through a one-dimensional distances, the simulations indicate convergence of the rock
wave propagation model coupled with the equivalent-linear and deep soil site peak motions with decreasing distance. In
representation of soil properties (Silva, 1991). Humphrey et the near field, the simulations produced higher motions than
al (1993) were able to produce a good fit to strong motions those predicted by the empirical model (1).

Seismological Research Letters Volume 68, Number 1 January/February 1997 65

 .5
....••.,
,.-....
l';))
...__... .2
4tw
.......
~
0 .1 •
.......,
...,

tj .05 Peri
~
.....,
~
PGJI
~
(.) .02
(.) 0.01
~
.01 0.1
~ 0.2
~ .005
a... M 8, rock M 8, shallow soil M 8, deep soil 0.3
• Simulations • Simulations • Simulations
.002 Empirical (1) Empirical ( 1) Empirical ( 1)
0.4
0.5
.001
10 20 50 1 00 200 500 10 20 50 100 200 50010 20 50 1 00 200 500 0.75
Distance (km) Di~tance (km) Distance (km) 1.0
1.5
• Figure 3. Comparison of PGA values predicted using regression model (1) and finite-fault stochastic model simulations. 2.0
The fitted empirical relationships are poorly constrained for interface earthquakes listed in Table 2. At distances 3.0
at small distances due to lack of data from multiple earth- greater than 30 to 40 km, the simulations compare well with
quakes. Therefore, the form of the regression model was the empirical interface model. At smaller distances, the sim-
modified to force convergence of the predictions for soil and ulations /predict higher motions than the empirical interface
rock motions at very small distances. The modified model is model for all three magnitudes. Also shown on Figure 5 are
given by: PGA values predicted by a shallow crustal attenuation rela-
tionship for rock motions from reverse faulting earthquakes
(Sadigh eta!., 1993). The near field simulations results are
consistent with the shallow crustal attenuation model. PGA
Figure 6 compares the PGA predictions from the atten- 0.07
uation models listed in Table 2 to those obtained by Crouse 0.1
(1991), Fukushima and Tanaka (1990), and Iai eta!. (1993)
+C5Z, +C8Zt +C9 H; +TJ; +Eii, 0.2
forM 6, 7, and 8 interface earthquakes. The Crouse (1991)
c; = C1 + C3C4 - c;c~ (2) soil model predicts PGA values that are very similar to those 0.3
obtained using the deep soil model listed in Table 2. The 0.4
c;=C3 +C6Z, Fukushima and Tanaka (1990) model predictions are consis-
0.5
C~ = C 4 +C7 Z, tent with the model developed in this study forM 6 events
but tend to give much lower results at large distances as the 0.75
When (2) was fit to the data, the resulting relationships pre- magnitude of the earthquake increases. These comparisons 1.0
dicted higher PGA values for rock in the near field than suggest that the differences between PGA attenuation of 1.5
obtained using (1), but lower soil PGA values. The reduction shallow crustal and interface earthquakes are significant pri-
marily for very large earthquakes. Unlike the other three 2.0
in near field soil amplitudes is likely due to the lack of near
field soil data. Therefore, we judged it'appropriate to use the attenuation relationships shown on Figure 6, the model pro- 3.0
soil attenuation model obtained by fitting model (1) and the posed by Iai et aL (1993) does not include the e[.f~t_£_f.D,f#r 4.0
rock attenuation model obtained by fitting model (2). The. ~-ei~ ..W~it.!lde ~t~at~Qn. Thus, while it predicts motions
consistent with the other models at magnitudes and dis-
y
selected attenuation models are listed in Table 2. Figure 4
tances within the bulk of the data, it does not match the M
compares the selected attenuation relationships to the
other models at small distances, underpredicting the other rrup
recorded PGA data.
The issue of near field motions was further examined by models at M 6 and greatly overpredicting the other models at H
conducting simulations of rock site motions forM 7, 8, and M8. Zr
8.5 interface events. Figure 5 compares the results of these The results of the regression analyses of the PGA data *St.
simulations to the PGA values predicted by the relationship indicated that intraslab earthquakes produce peak motions

66 Seismological Research Letters Volume 68, Number 1 January/February 1997

 TABLE 2
Attenuation Relationships for Horizontal Response Spectral Acceleration (5% Damping) for Subduction Earthquakes
For Rock
ln(y) = 0.2418 + 1.414M + C1 +C2 (10 -M) 3 +C3 ln(r,.,P + 1.7818e0· 554M) +0.00607 H +0.3846ZT
Standard Deviation = C 4 + C 5M

Period(s) ~ ~ ~ Cl Cs*
PGA 0.0 0.0 -2.552 1.45 -0.1
0.075 1.275 0.0 -2.707 1.45 -0.1
0.1 1.188 -o.0011 -2.655 1.45 -D.1
0.2 0.722 -Q.0027 -2.528 1.45 -0.1
0.3 0.246 -o.0036 -2.454 1.45 -D.1
0.4 -0.115 -o.0043 -2.401 1.45 -0.1
0.5 -Q.400 -o.0048 -2.360 1.45 -Q.1
0.75 -1.149 -o.0057 -2.286 1.45 -0.1
1.0 -1.736 -0.0064 -2.234 1.45 -D.1
1.5 -2.634 -0.0073 -2.160 1.50 -0.1
2.0 -3.328 -0.0080 -2.107 1.55 -0.1
3.0 -4.511 -0.0089 -2.033 1.65 -0.1
For Soil
ln(y) = -0.6687 + 1.438M + C1 +C2 (10 -M) 3 +C 3 !n( R + 1.097e0 ·617M) +0.00648H +0.364 VT
Standard Deviation= C4 + C 5M

Period(s) c1 ~ Ca C4* ~·
PGA 0.0 0.0 -2.329 1.45 -0.1
0.075 2.400 -0.0019 -2.697 1.45 -0.1
0.1 2.516 -0.0019 -2.697 1.45 -0.1
0.2 1.549 -0.0019 -2.464 1.45 -0.1
0.3 0.793 -o.0020 -2.327 1.45 -0.1
0.4 0.144 -0.0020 -2.230 1.45 -0.1
0.5 -Q.438 -0.0035 -2.140 1.45 -0.1
0.75 -1.704 -o.0048 -1.952 1.45 -0.1
1.0 -2.870 -0.0066 -1.785 1.45 -0.1
1.5 -5.101 -o.0114 -1.470 1.50 -0.1
2.0 -6.433 -o.0164 -1.290 1.55 -0.1
3.0 -6.672 -o.0221 -1.347 1.65 -D.1
4.0 -7.618 -o.0235 -1.272 1.65 -0.1
y =spectral acceleration in g
M =moment magnitude
rrup = closest distance to rupture (km)
H =depth (km)
Zr = source type, 0 for interface, 1 for intraslab
* Standard deviation for magnitudes greater than M 8 set equal to the value forM 8

Seismological Research Letters Volume 68, Number 1 January/February 1997 67

 •
•
•

0 0

• 0

Interface
M 5.3-5.7
0
t
Rock (33)
Deep Soil {27)
0
+
Interface
M 5.8-6.2
Rock (27)
Deep Soil (40)
0
Interface
M 6.3-6.7

t
Rock {23)
Deep Soil {42)
. \
\

0 '-
0 '
tJO\
0
0 ''
0 •
0 • '
.o,
Interface .'\t Interface
M 6.8-7.2 ' M 7.8-8.2
0 Rock (21) C Rock (16) 0 Rock (22)
+ Deep Soil (26) t Deep Soil (28) t Deep Soil {25)

100 200 50010 20 50 100 200 500 10 20 50 100 200 500

20 50

Distan ce (km) Distan ce (km) Distan ce (km)

.&. Figure 4. Comparison of PGA values predicted using attenuation models listed in Table 2
and the empirical data for interface earth-
line is the attenuation relationship
quakes. Numbers in parentheses give the average depth of earthquakes in each data subset. The solid
fit to deep soil site data and the dashed line is the relationship fit to the rock site data.

that are on average about 50 percent higher than those for directly to the calculated peak spectral ordinates may be
interface earthquakes for the same magnitude and distance. biased. We therefore developed relationships for response
Figure 7 compares the predicted PGA values for intraslab spectral amplification (SA/PGA). We followed the approach
earthquakes with the recorded data. of Youngs et a/. (1988) and included both magnitude and
distance effects on response spectral amplification. The spec-
ANALYSIS OF RESPONSE SPECTRA ORDINATES tral amplification relationship for any one spectral period is

The number of digitized and processed accelerograms for

subductio n zone earthquakes is only a subset of the PGA (3)
' database and is often limited to recordings with the strongest
shaking from each earthquake. Therefore, attenuation rela-
tionships for peak response spectral acceleration (SA) fitted

68 Seismological Research Letters Volume 68, Number 1 January/February 1997

 .5
•
.....-..
b)
--.....- .2 •
~ .1
0
·~
-;..,
tl .05
~
<l>
.......,
<l>
CJ .02
CJ
~
.01
~
tl
<l> .005 M 7, rock
a... • Simulations
M 8, rock M 8.5, rock
• ~.imulations • Simulations
.002 Interface Interface Interface
Crustal C. rustal Crustal
.001
10 20 50 100 200 50010 20 50 1 00 200 500 10 20 50 100 200 500
Distanc e (km) Distanc e (k1n) Distanc e (km)
.&. Figure 5. Comparison of PGA values predicted using the interface relationship for rock sites listed in Table
2 and finite-fault stochastic
model simulations for M 7, 8, and 8.5 events. Also shown are PGA values predicted using the shallow crustal
model of Sadigh eta/.
(1993).

The second term accounts for the magnitude scaling of ships listed in Table 2. We followed the standard convention
ground motions, and the third term for distance depen- and assumed that the individual peak motions are lognor-
dence. The coefficients a 1 and a2 are set equal to the coeffi- mally distributed. Youngs et al. (1988) found that the scatter
cients C4 and C5 of the appropriate PGA attenuation of peak acceleration data about the median attenuation rela-
relationship. The remaining variables have the same meaning tionship decreased with increasing magnitude . This effect
as those used in the PGA regression model. The coefficients has been reported in previous studies of crustal earthquakes
obtained at individual periods were then smoothed so that (e.g., Sadigh et ai., 1986; Abrahamso n, 1988). Youngs et aL.
the resulting spectral shapes are smooth over the full range of (1995) conducted a rigorous examinatio n of California
magnitude s and distances. Coefficients B2 and B were strong motion data using the random effects regression
3
found to be nearly linearly dependent on the log of spectral model and concluded that both inter-event and intra-event
period. Figure 8 shows the resulting spectral shapes for M componen ts of ground motion variability are magnitude
6.5, 7.5, and 8.5 events at distances of 50 and 200 km. The dependent. The need for such magnitude dependenc e in the
soil and rock spectral shapes are similar for smaller magni- subduction zone data set was investigated by fitting linear
tudes and then begin to diverge at larger magnitudes, with relationships of the form r = VI + VzM and (J' = v3 + V4M to
the soil spectral shape having more long-period motion, as the variance terms of (1) and (2). We found that, using the
one would expect. The spectral shapes developed in this likelihood ratio test (Seber and Wild, 1989), the null
study for soil site motions are similar to those derived from hypothesis that Vi = V4 = 0 can be rejected at the 1 percent
the analysis presented by Crouse (1991). significance level. We also found that the variance in the
The coefficients obtained by fitting (3) to spectral updated data set is larger than previously reported by Youngs
amplification values were then combined with the appropri- et ai. (1988). The resulting total variance for PGA was
ate attenuation relationships for PGA to produce attenua- approxima ted by a linear function ofM (Table 2).
tion relationships for 5% damped SA. These relationships The variance for SA was computed for the individual
are listed in Table 2. periods using the coefficients listed in Table 2 to define the
median attenuation relationships. The resulting estimates
GROUND MOTION VARIABILITY were somewhat lower than the values obtained for PGA.
Examinatio n of the analysis results indicates that the inter-
The remaining componen t of the attenuation relationships event componen ts of the variance were nearly zero for most
is an assessment of the variability of the peak motions of periods. This result is likely due to the limited number of
individual recordings about the median attenuation relation- earthquake s represented in the spectral ordinate data set.

Seismological Research Letters Volume 68, Number 1 January/February 1997 69

----------------------~-·-

.5

.2
•1 .... ....
·~

.." ..
. :-..._,
.05
• •'\
•
.02 • '\
.01
~ \
~ .005
Q,

.002
\
M 6 rock M 7 rock M 8 rock
.001

.5

.2

.1

.05

.02

.01
~
~ .005
Q,

.002
\
M 6 deep soil M 7 deep soil M 8 deep soil
.001
\
10 20 50 1 00 200 500 1 0 20 50 100 200 50010 20 50 100 200 500

Distance (km) Di,>tance (km) Distance (km)

.A. Figure 6. Comparison of PGA attenuation relationships for rock and deep soil sites. Data are from Figure 2. The points are the PGA
values from Figure 2 for the listed magnitude ±0.2 units.

Therefore, we increased the total variance for the spectral motions from subduction zone earthquakes will be larger
ordinates to account for an inter-event component of vari-- than those predicted using attenuation relationships for shal-
ance estimated from the peak acceleration data. The result-- low crustal earthquakes. The difference between these types
ing relationships are listed in Table 2. of earthquakes increases -as the size of the earthquake
increases. At small source-to-site distances, the empirical
DISCUSSION models developed in this study predict that the peak motions
from interface earthquakes are lower than those for shallow
The attenuation relationships developed in this study are crustal earthquakes. However, the near field data are very
considered appropriate for earthquakes of magnitude M 5 limited and numerical simulations indicate that peak
and greater and for distances to the rupture surface of 10 to motions may be similar to those predicted using shallow
500 km. The attenuation models indicate that for large crustal earthquake attenuation relationships. Therefore, we
events at large distances, one should expect that the peak suggest that one should consider estimates of peak motions

70 Seismological Research Letters Volume 68, Number 1 January/February 1997

 .5

-----
D)
...........- .2 • ••
~
0 0 1
• ''
''
•
·<->
....., '' '
d .05
~
.....,
~

~
••
<.) .02
<.)
0
~
.01 • \
\

~ Intra lab Intra lab

t'
\
~ .005 \
a., M 5.3-5. 7 M 5.8-6. 2 \
\
0 Rock (79) 0 Rock (7!i) \

.002
• Deep Soil (108) \
\
• Deep Sod ( 106)
\

.001
•
.5 ~
••
------
D) .2

.. ••
'---""

~ .1 '
0
·<-> ' ',
.....,
d .05
~
~
_,
~
<.)
<.)
.02
•
~
.01
~
d - Intra lab Intra lob
~ .005 lntrala b
a., M 6.8-7. :' M 7.3-7. 7
M 6.3-6. 7 Rock ( 100)
Rock ( 65) 0
Rock ( 49) 0
.002
0

• Deep Soil (85) • Deep Soil (90) • Deep Soil (100)

. 001 50 100 200 500

10 20 50 100 200 500 10 20 50 100 :-:Jo 5001 0 20

Dist ance (kin) Dist ance (km)

Dist ance (km)
b earth-
tion models listed in Table 2 and the empirical data for intrasla
A Figure 7. Comparison of PGA values predicted using attenua ship
akes in each data subset. The solid line is the attenuation relation
quakes. Numbers in parentheses give the average depth of earthqu
fit to the rock data.
fit to deep soil site data and the dashed line is the relationship
l
the intraslab motions would decrease with increasing spectra
in the near field of interface earthquakes using both ation relatio nships for SA for
period. Thus, the attenu
models developed in this study and those obtain ed using hat
intraslab earthquakes listed in Table 2 may be somew
attenu ation relationships for shallow crustal earthquakes. II
conservative at longer periods.
Because of the very limite d numb er of processed
2 are
intraslab recordings, the spectral shapes listed in Table
based primarily on interfa ce earthq uake data. The attenu a- REFERENCES
that
tion coefficients presented by Crouse (1991) indicate Abrahamson, N.A (1988). Statistical propenies of peak ground
accel-
earthq uake depth but long period 78,
PGA values increa se with erations recorded by the SMART! array, Bull. Sdsm. Soc. Am.,
se with increa sing earthq uake depth. If the
SA values decrea 26-41.
m for
larger PGA values for intraslab earthquakes are a result
of Abrahamson, N.A., and R.R. Youngs (1992). A stable algorith
analysis using the random effects model, Bull. Seism.
then on
higher stress drops for events ruptur ing oceanic crust,
regressi
differe nce betwe en interfa ce and Soc. Am., 82, 505-51 0.
one would expect that the

Volume 68, Number 1 January/February 1997 71

Seismological Research Letters
l:
I

This study - rock - - - This study -deep soil Crouse (1991)

5

•'
2

2§
~ .5
~
.2
.1 M 8.5, 50 km
M 6.5, 50 km
.05
5

2§
0... .5
'-.......
U§
.2

.1 . M 8.5, 200 km
M 6.5, 200 km M 7.5, 200 km
.05
.02 .05 .1 .2 .5 2 5 .02 .05 .1 .2 .5 2 5 .02 .05 .1 .2 .5 2 5

Period (sec) Period (sec) Period (sec)

.& Figure 8. Acceleration response spectral shapes (SNPGA, 5% damping) for rock and deep soil sites computed using the relationships
listed in Table 2. Also shown are spectral shapes computed using the attenuation relationships developed by Crouse (1991 ).

Boore, D.M., W.B. Joyner, and T.E. Fumal (1993). Estimallon of Japan, in Proc. Second Int. Conf on Microzonation for Safer Con-
response spectra and peak accelerations from western North struction, Research, and Application, II, 705-716.
American earthquakes, U.S. Geological Survey Open-File Report Joyner, W.B., and D.M. Boore (1981). Peak horizontal acceleration
93-509, 72 p. and velocity from strong-motion records including records from
Rrillinger, D.R., and H.K. Preisler (1985). Further analysis of the the 1979 Imperial ValJ.ey, California earthquake, BuU. Seism. Soc.
Joyner-Boore attenuation data, BuU. Seism. Soc. Am., 75 611- Am., 71, 2,011-2,038.
614. Joyner, W.B., and D.M. Boore (1993). Methods for regression analysis
Crouse, C.B. (1991). Ground-motion attenuation equations for earth- of strong-motion data, BuU. Seism. Soc. Am., 83, 469-487.
quakes on the Cascadia subduction zone, Earthquake Spectra, 7, Krinitz:ky, E.L., F.K. Chang, and O.W. Nuttli, references in Krinitz:ky,
210-236. E.L. (1987). Empirical relationships for earthquake ground
Crouse, C.B., Y.K. Vyas, and B.A. Schell (1988). Ground motions motions in Mexico City, in Proc. ASCE Conf: The Mexico Earth-
from subduction-zone earthquakes, BuU. Seism. Soc. Am., 78, 1- quake-1985, Factors Involved and Lmom Learnu/., Mexico City,
25. September 19-20, 1986.
Fukushima, Y., and T. Tanaka (1990}. A new attenuation relation for Molas, G.L., and E Yamazaki (1995). Artenuation of earthquake
peak horizontal acceleration of strong earthquake ground motion ground motions in Japan including deep focus events, Bull Srism.
in Japan, Bull Seism. Soc. Am., 80, 757-783. Soc. Am., 85, 1,343-1,358.
Hanks, T.C., and H. Kanamori (1979). A moment magnitude :scale,]. Sadigh, K., (1979). Ground motion characteristics for earthquakes
Geophys. Res., 84, 2,348-2,350. originating in subduction zones and in the western United States,
Humphrey, J.R., W.J. Silva, and R.R. Youngs (1993). Factors influenc- in Proc. Sixth Pan Amer. Conf, Lima, Peru.
ing site-specific ground motion estimates for the 1985 M 8.1 Sadigh, K., C.-Y. Chang, N.A Abrahamson, S.J. Chiou, and M.S.
Michoacan earthquake (abs.), Seism. Res. Lett., 64, 17. Power (1993). Specification of long-period ground motions:
Iai, S., Y. Matsunga, T. Morita, H. Sakurai, E. Kurata, and K. Mukai updated attenuation relationships for rock site conditions and
(1993). Attenuation of peak ground acceleration in Japan, inProc. adjustment factors for near-fault effects, in Proc. ATC-17-1 Semi-
Int. Workshop on Strong Motion Data, Menlo Park, California, nar on Seismic Isolation, Passive Energy Dissipation, and Active Con-
December 13-17, 2, 3-21. trol, March 11-12, San Francisco, California, 59-70.
Iwasaki, T., T. Kataymas, K. Kawashima, and M. Saeki (1978). Statisti- Sadigh, K., J .A. Egan, and R. R. Youngs ( 1986). Specification of ground
cal analysis of strong-motion acceleration records obtained in motion for seismic design oflong period structures (abs.), Earth-
quake Notes, 57, n. 1, 13. Relationships printed in W.B. Joyner

72 Seismological Research Letters Volume 68, Number 1 January/February 1997

 and·6.M. Boore (1988). Measurement, characterization, and pre- Youngs, R.R., S.-J. Chiou, WL. Silva, and J.R. Humphrey (1993).
- diction of strong ground motion, in Earthquake Engineering and Strong ground motion attenuation relationships for subduction
Soil Dynamics !!-&cent Advances in Ground Motion Evaluation, zone earthquakes based on empirical data and numerical model-
ASCE Geotechnical Special Publication 20, 43-102. ing (abs.), &ism. &s. Lett., 64, 18.
Seber, G.AF., and C.J. Wdd (1989). Nonlinear Regression, John Wiley Youngs, R.R., S.M. Day, and J.P. Stevens (1988). Near field motions on
and Sons, New York. rock for large subduction zone earthquakes, in Earthquake Engi-
Silva, W.J. (1991). Global characteristics and site geometry, in Proc. neering and Soil Dynamics !!-&cent Advances in Ground Motion
NSF!EPRI Workshop on Dynamic Soil Properties and Site Charac- Evaluation, ASCE Geotechnical Special Publication 20, 44 5-462.
terization, Electric Power Research Institute Research Project NP-
7337.
Silva, W.J., and C.L. Stark (1992). Source, path, and site ground Geomatrix Consultants
motion model for the 1989 M 6.9 Lorna Prieta earthquake, 100 Pine St., 1Oth Floor
research report prepared for the California Division of Mines and San Francisco, CA 94111
Geology. (R.R.Y., S.-].C.)
Tichelaar B.W., and L.J. Ruff(l993 ). Depth of seismic coupling along
subduction zones,] Geophys. &s., 98, 2,017-2,037.
Vyas, Y.K., C.B. Crouse, and B.A Schell (1984). Ground motion Pacific Engineering flnd Analysis
attenuation equations for Benioff zone earthquakes offshore 311 Pomona St.
Alaska (abs.), Earthquake Notes, 55 (1), 17. El Cerrito, CA 94530
Wyss, M., and R.E. Habermann (1982). Conversion of mb to M for
5 (W.].S.)
estimating the recurrence time of large earthquakes, Bull Seism.
Soc. Am., 72, 1,651-1,66 2.
Youngs, R.R., N.A. Abrahamson, F. Makdisi, and K. Sadigh (1995). Lahontan GeoScience Inc.
Magnitude dependent dispersion in peak ground acceleration, 21380 Castle Peak Rd.
Bull. Seism. Soc. Am., 85, 1,161-1,176. Reno, NV89511 -7119
(JR. H)

Seismological Research Letters Volume 68, Number 1 January/February 1997 73