A S T IE H N I I ~ I D E M I I i I [ I I I l l

SEISMOLOGICAL
IRESEARCH~

Attenuation of Strong Ground Motion in

Peninsular India
R. N. lyengar and S. T. G. Raghu Kanth
Dept. of Civil Engineering, Indian Institute of Science, Bangalore

ABSTRACT the Khillari (29 September 1993), Jabalpur (22 May 1997),
and Kutch (26 January 2001) shocks. But the available quan-
Earthquakes in India occur in the plate-boundary region of tified information is so sparse, engineers presently face a
the Himalayas as well as in the intraplate region of peninsular daunting problem in estimating ground-motion levels for
India (PI). Devastating events have occurred in PI in the future events in PI. The present paper is motivated by the
recent past, which is a warning about the possibility of such need to have a simple approach to understand attenuation in
earthquakes in the future. But very limited recorded data are PI from the engineering point of view.
available about ground motion in PI for engineers to rely First, the database available for PI is briefly reviewed. PI
upon. The present paper, after a review of available data, is not homogenous with respect to its seismogenic character.
develops an attenuation relationship based on a statistically The available strong-motion accelerograph (SMA) data do
simulated seismological model. The proposed equation for not cover the entire region and lack multiple SMA recordings
peak ground acceleration (PGA), under bed rock conditions, of individual events. Hence an empirical equation obtained
is of the form solely with the help of the available incomplete SMA data will
be unreliable. Here, this difficulty is circumvented by adopt-
ln(PGA/g) = c 1 + c 2 ( M - 6 ) + s 6) 2 - In(R) - c4R + lne, (1) ing a seismological model for synthetic generation of PGA
values following Boore (1983). The present study includes
with c1 - - 1 . 6 8 5 8 , c2 =0.9241, c3 =-0.0760, c4= 0.0057, regional differences of quality factor within PI and effect of
and o'(lnt3) - 0.4648. Correction factors for other site condi- uncertainties in stress drop, radiation coefficient, cut-off fie-
tions are also computed. In the absence of a robust database quency, and focal depth. With the help of a large synthetic
of strong-motion records, seismological modeling is a ratio- database, an attenuation relation is obtained by a two-way
nal alternative until sufficient instrumental records become stratification approach (Joyner and Boore, 1981). The pro-
available in PI. It is observed that attenuation of strong posed attenuation relation is found to match favorably with
motion in PI is similar to that in other intraplate regions of the estimated PGA values of the Kutch earthquake of 2001.
the world.
PENINSULAR INDIA
INTRODUCTION
The region of the Indian subcontinent south of 24~ latitude
India has faced several devastating earthquakes in the past. is taken here as peninsular India (PI). This landmass is far
The largest of these have originated in the Himalayan plate- away from the Himalayan collision zone, the well known
boundary region, which remains a region of great scientific boundary between the Indian and Asian plates. Nonetheless,
and engineering interest. Not surprisingly, considerable data it is recognized that Cambay and Rann of Kutch in Gujarat
and earthquake related literature are available about the are among the active regions of India. Apart from this region
northern part of India. On the other hand, very little seismo- and the Andaman-Nicobar Islands, the remaining part of
logical information is available about peninsular India (PI), continental PI has reliably experienced some 400 earthquakes
which is taken here as south of 24~ latitude. This situation in a period of 600 years. This number would be much larger
is changing, in response to three recent devastating events: if all instrumentally recorded shocks of small magnitudes

530 SeismologicalResearchLetters Volume75, Number4 July/August2004

 TABLE 1
Earthquakes of Peninsular India
Epicentral
Focal Mag. Intensity
Location Date Epicenter (N/E) Depth(km) (Mw) (Io) Reference
C0imbat0re 7 February 1900 10.8~ ~ -- 6.1 VIII Kaila et aL (1978)
Vijayanagaram 17 April 1917 18.0~ ~ -- 5.5 VII IMD
Anjar 21 July 1956 23.0~ ~ -- 6.1 IX Kaila et al. (1978)
Ong01e 13 October 1959 15.6~ ~ ~ 5.0 VII Chandra (1977)
Ong01e 27 March 1967 15.6~ ~ ~ 5.8 VII IMD
K0yna 13 September 1967 17.4~ ~ -- 5.8 VIII Guha et aL (1970)
K0yna 11 December 1967 17.5~ ~ 10 6.5 VIII-IX Guha et aL (1970)
Bhadrachalam 13 April 1969 17.9~ ~ 25 5.7 VII Kaila et al. (1978)
Broach 23 March 1970 21.70/72.9~ 8 5.8 VII Kaila et al. (1978)
C0imbat0re 29 July 1972 11.0~ ~ -- 5.0 VI IMD
Shim0ga 12 May 1975 13.8~ ~ 10 5.0 V Kaila et aL (1978)
Krishnagiri 20 March 1984 12.58~ ~ 6 4.5 VI Guha et aL (1993)
Idukki 7 June 1988 9.81 ~ o 5 4.5 V Rast0gi et aL(1995)
Khillari 29 September 1993 18.06~ ~ 7 6.2 VIII Narula et aL (2000)
B0naigarh 29 March 1995 21.660/84.59~ 21 4.7 V Narula et aL (2000)
Jabalpur 21 May 1997 23.0o/80.0~ 35 5.8 VIII Narula et al. (2000)
Bhuj, Kutch 26 January 2001 23.40/70.28~ 24 7.7 Xl IMD
Bhuj, Kutch 28 January 2001 23.61~ ~ 15 5.7 VII IMD
IMD: India Meteorological Department, New Delhi, India.

were also included. A list of damaging earthquakes of engi- Seeber et al. (1999) studied the seismicity of PI with particu-
neering importance that have occurred in PI in the last 100 lar reference to Maharashtra. They concluded that between
years is presented in Table 1. 1960 and 1990 the seismicity of PI showed a threefold
The seismicity of PI from a seismological perspective has increase. This was the period during which industrial devel-
been discussed in the past notably by Chandra (1977) and opment also increased severalfold in PI. Thus, engineers must
Rao and Rao (1984). A catalog of PI earthquakes of magni- recognize that the looming seismic risk to man-made struc-
tude greater than 3 was compiled by Guha and Basu (1993). tures in PI is more than what was previously believed.
It is generally held that seismic activity is greater at the inter-
sections of the Dharwad, Aravali, and Singhbhum protocon- GROUND-MOTION DATABASE
tinents, which together constitute PI (Figure 1A). With data
available up to 1984, Rao and Rao (1984) fitted the fre- This is a brief review of strong-motion data available for PI.
quency-magnitude relationship The only region with SMA data is the Koyna-Warna region
of western India (Figure 1B). The earliest available PGA value
Log10 N = a - b M (2) for PI is from the Koyna earthquake record of 11 December
1967. After this, a large number of records of smaller magni-
to get a = 4.4 and b = 0.85 for PI. They also demonstrated tudes were obtained in the Koyna region. This set of data,
that the interval 1870-1920 was a period of quiescence, taken from the reports of Gupta et al. (1992), is presented in
whereas prior to and after this time window PI showed higher Table 2. For the mainshock of the Khillari earthquake of
levels of seismic activity. It may be noted that the Indian code 1993, no near-source ground motion records are available.
on earthquake-resistant design of structures, IS-1893, pre- SMA records have been obtained by Baumbach et al. (1994)
sents a zonation map of India which was revised after the for a few aftershocks of this event, however. Three such values
Koyna and Khillari earthquakes. The revision has been to are given in Table 3. A few instrumental velocity records
upgrade parts of PI from zero and low levels of seismic hazard within epicentral distances of 300 km are available for the
to higher levels. Thus, there is a consensus among scientists Jabalpur earthquake of 1997 (Singh et al., 1999). Similarly, a
and engineers that PI is passing through a peak in its activity. few data are available for the main event and aftershocks of

Seismological Research Letters July/August2004 Volume 75, Number4 531

 (A) 24
Kutcb Aravali
I9
S
,. T0

s ( ~ Jabalpur s Sin~lhbum
2001! Protocontineni
9
f
~ "* 1997., r Protocontinent
s
~o~'ch"
1970
"
,,,s
f
'
" 9
Bonaigarh r
r . bl

20 .......
i0
9 I J ~%
I. . . . . . . . . . . . . . . .
"~,. ~ ........
!

:! |
f t
"4 I i) I
S
Q K~i~a~i Bhadrachalar~ ~. I
I Koyna 199~ ( ~ 1969 J ( ~ I
I ~ 1967
!

Z16
I | Dharwar "~. J
. %
.

O ....... I., . . . . . . . P&btb~Ohtihertt"

!

I
l tll~ o M <4
II ~ ~ Sh."imoga
tl 9.75 0
9 I o 4<M ~5
i I W.
12 : ........ ~.
|
............ .Q... 5.<.M W...~.6....
%
I
%
~ i (.;oimbL
coirnl
I
6< M W ' -~7
~1900 I

' 0 Idukki , @ M>7!

! w
S !
!

8 1 ,
I
68 72 76 " - - "80" " 84 88
o E

(B)
NESTERN-CE!NTRAL
I REGION

........................................................... j ........
20
%

9 %:

KOYNA- I " ,~
WARNA I " ",
II
!

O
z16 ......... ~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
I.
. ~,

SOUTHERN
REGION

12

8 ~
68 72 76 80 84 88
o E

A Figure 1. (A) Peninsular India. (B) Three subregions of peninsular India.

532 Seismological Research Letters Volume 75, Number4 July/August2004

 TABLE 2
Instrumental PGA for Koyna-Warna Region
Epicentral Focal Depth RecordedPGA EstimatedPGA(g)
Epicenter (N/E) Date Mw Distance(km) (km) (g) (Eq. 11)
17.50~ ~ 13 September1967 5.6 13 3 0.1640 0.2651
17.48~ ~ 13 September1967 4.3 11 5 0.0200 0.0726
17.50~ ~ 10 December1967 6.5 13 10 0.4860 0.4827
17.28~ ~ 12 December 1967 4.5 14 13 0.0360 0.0546
17.29~ ~ 13 December1967 4.4 12 15 0.0410 0.0483
17.35~ ~ 24 December1967 4.8 7 20 0.0500 0.0676
17.35~ ~ 24 December1967 4.8 7 20 0.0350 0.0676
17.33~ ~ 4 March 1968 4.0 4 10 0.0190 0.0576
17.36~ ~ 4 March 1968 4.0 9 10 0.0090 0.0450
17.36~ ~ 1 January 1970 4.1 9 11 0.0130 0.0479
17.32~ ~ 27 May 1970 4.2 11 3 0.0698 0.0688
17.48~ ~ 26 September1970 4.2 11 13 0.0420 0.0439
17.36~ ~ 17 February 1974 4.5 17 19 0.0214 0.0387
17.49~ ~ 29 July 1974 4.1 8 24 0.0360 0.0244
17.27~ ~ 2 September 1980 4.1 18 13 0.0293 0.0286
17.23~ ~ 2 September1980 4.1 18 13 0.0110 0.0286
17.23~ ~ 20 September1980 4.5 21 8 0.0310 0.0450
17.20~ ~ 20 September1980 4.7 17 8 0.0220 0.0697
17.25~ ~ 20 September1980 4.7 17 8 0.0219 0.0697
17.24~ ~ 25 April 1982 4.1 18 13 0.0293 0.0286
17.24~ ~ 25 April 1982 4.1 18 13 0.0110 0.0286
17.35~ ~ 12 March 1995 4.5 20 10 0.0130 0.0452
17.32~ ~ 13 March 1995 4.2 25 10 0.0055 0.0255

TABLE 3
Instrumental PGA for Western-Central Region
Epicentral Focal Depth RecordedPGA EstimatedPGA(g)
Epicenter (N/E) Date Mw Distance(km) (km) (g) (Eq. 12)
17.93~ ~ 8 October 1993 4.3 21 5 0.0640 0.0373
17.93~ ~ 8 October 1993 4.3 11 5 0.0103 0.0709
17.93~ ~ 8 October 1993 4.3 13 5 0.0655 0.0608
23.8o/80.06~ 21 May 1997 5.8 237 36 0.0125 0.0043
23.8o/80.06~ 21 May 1997 5.8 271 36 0.0086 0.0030
23.61~176 28 January 2001 5.7 101 15 0.0079 0.0218
23.61~176 28 January 2001 5.7 249 15 0.0017 0.0035

the Kutch earthquake of 2001. PGA values obtained by dif- 1995; Hwang and Huo, 1997; Toro et aL, 1997). Singh et aL
ferentiating the instrumental velocity records of the Jabalpur (1999) used a seismological model for estimating ground
earthquake and Kutch aftershock (Singh et al., 2003) are also motion in parts of PI, but no specific attenuation equation
shown in Table 3. The moment magnitudes (Mw) of events has been proposed by them. The theory and application of
for which instrumental data are available in PI varies from 4 seismological models for estimating ground motion has been
to 7.7. The database does not have instrumental PGA values discussed in detail by Boore (1983, 2003). Briefly, the Fourier
in all distance ranges of engineering importance, however, amplitude spectrum of acceleration at bedrock is expressed as
and hence can not be used for empirical attenuation studies.
A(f) = C[S( f)]D( f ) P ( f ) (3)
SEISMOLOGICAL MODEL
where S ( f ) is the source spectral function, D(f) is the
In regions lacking strong-motion data, seismological models diminution function characterizing the attenuation, P(f) is
(Boore, 1983) are viable alternatives and are used worldwide a filter to shape acceleration amplitudes beyond a high cut-off
for deriving attenuation relationships (Atkinson and Boore,

Seismological ResearchLetters July/August2004 V01ume75,Number4 533

frequency fro, and C is a scaling factor. In the present study, Singh et aL (1999) derived Q as 5 0 8 f ~ from the broad-
the single corner-frequency model band records of the Jabalpur earthquake, for the western-cen-
tral parts of PI. These three regions which make up PI are
shown in Figure 1B. Thus to develop an attenuation relation
S(f)- (2 af)2 M~ (4) applicable for the whole of PI, the above quality, factors which
l+(f) 2 broadly represent three different regions within PI have to be
considered.
The seismological model is implemented in the time
domain in each region by using the method of Boore (1983,
of Brune (1970) is used, where the corner frequency f~, the 2003). The simulation procedure essentially consists of three
seismic moment M 0, and the stress drop A a are related steps. First, a Gaussian stationary random process sample of
through strong ground-motion duration (Boore and Atkinson 1987),

- 49x10, / 0/1' . (5)

1
T = ~ +0.05R, (9)

Here the shear-wave velocity V~in the source region is taken is simulated. The sample is windowed by multiplying it with
as 3.6 km/s. The diminution function D ( f ) is defined as the modulating function of Saragoni and Hart (1974) and is
Fourier-transformed into the frequency domain. The Fourier
amplitude spectrum is normalized by the square root of the
(6) mean square amplitude spectrum and multiplied by the tar-
D( f ) - G exp VsQ( f ) , get spectrum of Equation 3, derived from the seismological
model. This is transformed back into the time domain to
in which G refers to the geometric attenuation and the other generate a sample of acceleration time history. This way an
term to anelastic attenuation. In this equation Q is the quality ensemble of acceleration time histories is simulated.
factor of the region. The high-cut filter in the seismological The PGA samples obtained are based on a given set of
model is given by model parameters, which are themselves uncertain. Thus the
sample PGA values from a suite of such simulations may still
[ 1'2 not reflect all the variability observed in real ground motion.
To account for this, the four model parameters, namely stress
P(f'fm)- 1 + --~m (7) drop, focal depth, fm, and the radiation coefficient, are treated
as random variables, distributed uniformly about a mean
value. The stress drop is taken to vary between 100-300 bars
where fm controls the high-frequency fall of the spectrum. (Singh et al., 1999). From Table 1, it is seen that the focal
The scaling factor C is depth in PI has varied in the past around 10 km. With this in
mind, the focal depth is taken as a uniform random variable
in the range 5-15 km. The cut-off frequency, based on past
- (8)
SMA data, is taken in the interval 20-50 Hz. The range of
4 rcp Vs3 the S-wave radiation coefficient is taken as 0.48-0.64 (Boore
and Boatwright, 1984). It may be noted here that Hwang and
where (R0~) is the radiation coefficient averaged over an Huo (1997) also considered uncertainties in model parame-
appropnate range of azimuths and take-off angles. The coef- ters for deriving synthetic attenuation relationships for the
ficient a/2 in the above equation arises as the product of the central and eastern United States (CEUS). PGA values are
flee-surface amplification and partitioning of energy in simulated for moment magnitudes ranging from 4 to 8 with
orthogonal directions. an increment of 0.5 units. Epicentral distance is varied in
Following the work of Singh et al. (1999), the geometri- intervals of In(r) = 0.13, where r stands for the epicentral dis-
cal attenuation term G for the Indian shield region is taken to tance. The random vibration code written by Boore (1996) is
be equal to I / R for R < 100 km and equal to 1/(10 fiR ) for used for generating the synthetic data. The magnitude and
R > 100 km. For PI, we have three studies reported in the lit- range of epicentral distance chosen in each case are shown in
erature for finding the Q values. Mandal and Rastogi (1998) Table 4. A lower limit on the epicentral distance is imposed
have studied the Koyna-Warna region, which has a large num- since the seismological model uses a point-source assumption
ber of instrumental recordings, to arrive at Q = 169f .77. Rao (Kriniztsky et al., 1993). The number of distance samples
et al. (1998) used strong-motion records of small-magnitude considered for each magnitude is also shown in Table 4. In all
earthquakes and estimated Q value to be 4 6 0 f ~ for the there are 101 pairs of magnitudes and distances. For each
southern part of PI, excluding the Koyna-Warna region. magnitude, 100 sets of seismic parameters are generated,

534 SeismologicalResearchLetters Volume75, Number4 July/August2004

 TABLE 4 Southern Region:
Ranges of Epicentral Distance cI = 1.7816; c2 = 0.9205; c3 =-0.0673; c4 = 0.0035;
o(lne) = 0.3136. (13)
Number of
Epicentral Distance Distance
Typical attenuation curves for M w 6 are presented in Figure 2
Moment Magnitude (km) Samples
for all three subregions of PI. Mandal and Rastogi (1998)
4 1-300 20 have previously observed that attenuation in Koyna-Warna
4.5 1-300 20 region is faster, similar to what is expected in a tectonically
5.0 5-300 14 active region. The southern region shows slow attenuation, as
5.5 15-300 10 in a stable continental region. The western-central region
6.0 25-300 9 appears to lie between these two patterns. In the present
6.5 35-300 8 investigation PI has been represented by three regions with
7.0 40-300 7 different Q factors. Equations 11, 12, and 13, for all practical
7.5 45-300 7 purposes, are very similar, however. Hence, it would be con-
8.0 60-300 6 venient and also fairly accurate to have a single composite for-
mula for PI. The three regions~Koyna-Warna (84,948 km2),
where each set of seismic parameters defines a synthetic earth- southern region (3,39,793 km2), and western-central region
quake. Thus, the database consists of 10,100 PGA samples (4,24,742 km2)~cover PI nearly in the ratio 1:4:5. With this
from 900 earthquakes. This synthetic database is generated in mind, flesh PGA samples have been selected from the three
separately for the Koyna-Warna, western-central, and south- regional populations in the above ratio to create a new syn-
ern regions of PI using their respective quality factors. thetic database for PI. This contains 10,100 samples as before,
covering the same magnitude and distance ranges. For the
ATTENUATION proposed attenuation relation of Equation 10, the following
parameters are obtained by stratified regression for peninsular
A review of some of the general attenuation relations and the India:
methods appropriate for regression analysis have been
recently presented by Campbell (2003). For PI, a model is 6"1 = 1.6858; c2 = 0.9241; c3 =-0.0760; c4 = 0.0057;
proposed that accounts for geometrical spreading and anelas- and o'0ne) = 0.4648. (14)
tic attenuation similar to the equations derived for eastern
North America (ENA) and other intraplate regions. The cho- The above attenuation relation is valid for hard rock exposed
sen attenuation relation has the form on the surface, with Vs nearly equal to 3.6 km/s. For other site
conditions the results of the above equation will have to be
lny = 6"1 + 6"2(M - 6) + c 3 ( M - 6) 2 - l n R - c4R + lne (10) corrected as discussed below.

where y, M, and R refer to PGA(g), moment magnitude, and VALIDATION

hypocentral distance, respectively. Since PGA is known to be
distributed nearly as a lognormal random variable, In y would It has already been pointed out that the available SMA data in
be normally distributed with the average of ln/3 being almost PI do not cover all relevant magnitude and distance ranges.
zero. Hence with ~3= 1, Equation 10 represents a 50 percen- Hence, it would be relevant to see how the above synthetic
tile or median-level hazard estimation formula for PGA. A attenuation relation matches with available observations. In
two-step stratified regression analysis originally proposed by Table 2, the predictions of Equations 10 and 11 are compared
Joyner and Boore (1981) is carried out on the generated syn- with observed PGA data on hard rock in the Koyna-Warna
thetic data to obtain the parameters of the attenuation equa- region. Similarly in Table 3, some comparisons are presented
tion for bedrock condition (Vs = 3.6 km/s) as for the western-central region. In both these cases the sites are
hard rock with high Vs values and hence no corrections are
Koyna-Warna Region: applied.
cI = 1.7615; c2 = 0.9325; c3 =-0.0706; c4 = 0.0086; Another comparison of the new attenuation relation is
o'0ne) = 0.3292; (11) presented in Figure 3 for the Kutch earthquake of 26 January
2001. No instrumental observations are available for this
Western-central Region: event in the near-source region except for a few readings from
6"1 1.7236 ; c2 = 0.9453; c3 =-0.0740; c4 = 0.0064;
=
spectral response recorders (SRR). These recorders give the
o'(lnE) = 0.3439; and (12) peak response of a simple harmonic oscillator, tuned to a par-
ticular frequency and damping value. In the absence of direct
SMA recordings, the above data can be used for estimating
PGA. This provides an opportunity to verify the new a t t e n u -

Seismological ResearchLetters July/August2004 Volume75,Number4 535

 ,. , , , , , , , ! ' ,. , , , [ ,

, . : : : : : : ~oyna-warna
i i ! ~ s~176
Western-Central

i . . . . . . . . . . . . . . . . . . . .
9 i . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i ~i..... .... . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . : . . . . : .: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . : . . . . . : . ~ . . . . . . . . . . . . . . . . . . . .

. . . .
. . . .
. .

10 -~ . . . . . . . .

: . . . . . : .:

. . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . :. . . . . . .. .: . : . . . . . . . . . . . . . . . . . . \ .

. . . .
. . . . . . . . .

. . . .
. . . .
. . . .

Q ,

i i!i i i"i\
iiii i ! \

i ii! : ::: \
10 -;z ~ ~ ~ ~ ~ : ~ ~ ~ ~ ~ ~ ! ~ : ~ ~
.................................... i .i..i.; ..................................... i..!..i.i ......... -~.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ; . . i . . : . ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . i . . ; . ; . . . . . . . . . \ . .

........................... i ........... i i i i ......................................... i i ! i ............. \

9 ~ ~ ~ ~ ~ ~ ~

. . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . . . . . . , . , . . . . : . : . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . ~ . . . . . . , . ~ . . . . . . . . . . . . ..

9 i ii! . . . . . \
i i i i ii! i i !ii \
. . . . . . . . . .

9 . . . . . . . . . .

10 ~ 10 ~ 10 2
Epicentral Distance (km)

A Figure 2. Attenuation in peninsular India: M,, 6. ED = 10 km.

- 1 - 1 [ 1 I ,

100 ,"1; .................. .................... : ............... [] PGA(SRR) ~-

"~,) .:-~- ................ :.................. ! ............... -~ PGA (Cramer & Kumar (2003)) I-
"tJL il ................ i.................. : ............... - Western-Central Region I
" : .................. i.................. ! ................ PI I-

........ i~ .................................. ! .................. i .................. ! ..................

. _ . _ . . . . . . . . . . . . ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

\ , ,

.
(.9 1 ................................................... : ............. ~---" .................

[]

. . . . . . . . , . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . ~ . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . ,. . . . . . . . . . . . . .

10 -2 ~

so lOO 1so 200 250 300

Epicentral Distance (km)

,A, Figure 3. Analytical attenuation relation and PGA values of Kutch earthquake (B-C site condition).

536 Seismological Research Letters Volume 75, Number4 July/August2004

 TABLE 5
PGA Estimates from SRR Data (Kutch Region)
77= 0.05 77=0.1
Epicentral T= 0.4s T= 0.75s T= 0.4s T= 0.75s PGA ( g) PGA (g) B-C
Distance SRR SRR SRR SRR (Eqs. 15, Site (Cramer PGA (g) B-C
Station (km) S,(g) S,(O) s,(g) Sa(g) 16) and Kumar) site (authors)
Anjar 44 1.62 0.70 0.91 0.69 0.62 0.58 0.55
Kandla 53 0.86 0.57 0.65 0.53 0.41 0.33 0.38
Niruna 97 0.81 0.65 0.76 0.55 0.44 0.24 0.33
Naliya 147 0.72 0.22 0.69 0.21 0.40 0.22 0.29
Khambaliya 150 O.18 0.07 0.09 0.04 0.06 0.07 0.06
Jamjodhpur 166 0.22 O.15 O.14 0.05 0.07 0.09 0.07
Dwaraka 188 0.21 ~ 0.18 0.11 0.11 0.06 0.08
Porbander 206 O.19 0.25 O.15 0.21 0.10 0.05 0.07
Junagarh 216 O.19 0.06 O.10 0.05 0.07 0.08 0.07
Amreli 225 0.09 0.07 0.07 0.05 0.05 0.04 0.05
Ahmedabad 238 0.29 0.23 0.24 0.19 0.15 0.08 0.10
Cambay 266 0.49 0.04 0.29 0.04 0.19 0.14 0.14
Anand 288 0.14 0.06 0.12 0.05 0.08 0.04 0.06

ation relation for a strong earthquake in PI. In Table 5 the coefficients given by Joyner and Boore (2000). This set of
SRR values at thirteen stations as reported by the University adjusted PGA values of the Kutch event with M w 7.7 is also
of Roorkee (2001) are shown. To convert the SRR values into shown in Table 5. In Figure 3, attenuation of PGA during the
PGA values, first a multivariate regression between recorded event, as estimated by Cramer and Kumar (2003) and Equa-
PGA values of 240 sample strong-motion accelerograms of tion 15, is compared with the analytical results of Equations
the PEER (http://www.peer.berketey.edu) global database and 10 and 12, as corrected for B-C type soil condition. It is
their corresponding true Sa(r I, Tn) is performed. Here ,.,cais observed that the empirical equation is able to estimate the
the spectral acceleration, 77is the viscous damping coefficient, PGA values reasonably well.
and T, is the natural period of the oscillator. The regression
equation is of the form DISCUSSION
In the absence of sufficient past recorded data, the use of ana-
In(PGA) = a I + a2 In[S,(0.05,0.4)]+ a, 1n[S,(0.05,0.75)] (15) lytical models is the next best approach to estimate strong
+a4 In[ Sa(0.1, 0.4)1 + a4 In[ Sa(0.1, 0.75)1 + In& ground motion. Besides developing a model attenuation rela-
tion for PI, some limited validation of the model has also
been presented here. It is seen that for the Koyna earthquake
For the global data one finds
of 11 December 1967, the present estimate matches well with
the recorded value. In Table 2, the mean and standard devia-
a 1 =-0.5158; a 2 = 0.25; a 3 =-0.2488; a 4 = 0.8586;
tions of the differences between the estimated and observed
a 5 = 0.2922; and o(lnb') = 0.3429. (16) values are-0.023 and 0.025, respectively. In Table 3, it is
found that the mean error is nearly zero, with the standard
The estimated PGA values obtained for the Kutch earth- deviation being 0.027. For the 5.8 magnitude Jabalpur earth-
quake by the above multivariate regression are reported in quake, however, the predicted values at long distances are
Table 5. Previously Kumar et all. (2001) estimated the PGA lower than the recorded values. This perhaps can be attrib-
values using only S a (0.05, 0.4) data. The estimated PGA val- uted to the high stress drop of 420 bars associated with this
ues refer to recorders on different soil conditions and hence event (Singh et al., 1999). The comparison of the Kutch
cannot yet be compared directly with the derived attenuation earthquake PGA data with the model attenuation is also
relation. To circumvent this difficulty, following Cramer and favorable, with seven observations lying in the mean plus/
Kumar (2003) the soil conditions at the thirteen stations are minus standard deviation interval.
classified as B-C, C, and D sites (BSSC, 2001), and PGA val- It is quite well known that there can be considerable vari-
ues are estimated for firm rock (B-C) conditions using the ation in local site conditions, and hence the surface-level PGA

Seismological ResearchLetters July/August2004 Volume75, Number 4 537

value can be different from the exposed bedrock value. To conditions are shown for an event with M w 7.7. It is seen that
account for this it is expedient to adopt a classification scheme the analytical estimates compare favorably with the observed
as per NEHRP (BSSC, 2001), which groups sites based on PGA values of the Kutch earthquake, as explained previously.
the average shear-wave velocity (1/30) in the top 30 meters. The eastern part of North America (ENA) and PI share
Following this approach, correction factors to Equations similarities in that both are intraplate regions. With this in
10-14 have been found for A, B, and B-C type sites. To mind, a comparison of the present formula with a few other
describe typical site velocity profiles, the examples given by such relations for ENA (Atkinson and Boore, 1995; Hwang
Hwang et al. (1997) for A (1.5 < V30 < 3.6 km/s) and B and Huo, 1997; Toro et al., 1997) is shown in Figure 4. In the
(0.76 < 1/30 < 1.5 km/s) sites have been used. For firm-rock same figure, the attenuation relationship derived from world-
sites, which are between sites C and B, denoted as B-C types, wide intraplate earthquake data (Dahle et al., 1990) is also
the sample profile of Frankel et al. (1996) with shown. It is seen that all the results are qualitatively similar
V30 = 0.76 km/s has been used. Site-dependent PGA values Ys but show some scatter. It is interesting to observe that the
have been again simulated for these soil profiles, using the sto- derived attenuation relationship for PI matches well with
chastic seismological model of Boore (1996, 2003). From ENA ground-motion relations. Such a comparison between
these simulated results, a relation between Ys and y of Equa- Kutch PGA data and ENA attenuation was also observed by
tion 10 at the median level can be expressed in the form Cramer and Kumar (2003). In view of the above favorable
Ys = yFs. The correction coefficient F s is found to be 1.429 for points, it is proposed that Equations 10 and 14, suitably cor-
A sites, 1.643 for B sites, and 2.297 for B-C sites in PI. The rected for A, B, and B-C type sites, may be used for estimat-
standard error in the attenuation incorporating the site effects ing PGA in PI until more instrumental data are recorded.
would be o'(lneA) = 0.465, o'(lneB) = 0.467, and
cr(lneBc) = 0.469. SUMMARY
For other types of site conditions, such as C, D, and E,
further site-specific soil amplification studies are needed to One of the key parameters engineers look for is the PGA at a
arrive at surface-level PGA values from the attenuation for- chosen site for the design-basis earthquake. This is region-
mula derived here. In Figure 3, results of the composite atten- specific and hence every effort should be made to understand
uation relation of PI corrected for firm-rock (B-C sites)

............ : ...... : .... : : : i i i i ........... : ...... : .... : + Hwang&Huo(1997)

" ~ " ! ! " i ! ! ! ........... ! ...... ! .... !-- --x-- Atkinson&Boore(1995)
- ' ~ - . ~ ~ . !.! ........... ! ...... ! .... !-. ---- Dahle(1990)
i .......... :_ _ _ ~ ~ ~ . . . 9 . --0- Toro et al (1997)
~ ~ ; ...... ; .... ! : . : i ~ ' x ," ," x: , . ...........
. ! ...... ! .... ! -e- PI

...... i .... .... ......

.....

1 0 -1

I1.

1 0 -2

10 ~ 101 102
Epicentral Distance (km)

A Figure 4. Comparison of SCR attenuation curves: Mw 6, F.D = 10 kin.

538 Seismological Research Letters Volume 75, Number4 July/August2004

how peak ground parameters attenuate with distance for Baumbach, M., H. Grosser, H. G. Schmidt, A. Paulat, A. Rietbrock, C.
Indian earthquakes. At present no specific relation is available V. Ramakrishna Rao, E Solomon Raju, D. Sarkar, and I. Mohan
(1994). Study of foreshocks and aftershocks of the intraplate Latur
for engineers to use in PI, which is the intraplate region south
earthquake of September 30, 1993, India, in H. K. Gupta (editor),
of 24~ latitude. The strong-motion data available for PI are Latur Earthquake, Geological Society of India Memoir 35, 33-63.
very sparse and are confined to a limited region. For the deva- Boore, D. M. (1983). Stochastic simulation of high-frequency ground
stating Kutch earthquake of 26 January 2001 only SRR data motions based on seismological models of the radiated spectra,
are available, which at best can help in statistically estimating Bulletin of the SeismologicalSociety ofAmerica 73,1,865-1,894.
Boore, D. M. (1996). SMSIM: Fortran Programfor Simulating Ground
PGA values. Additionally, there have been major earthquakes
Motions from Earthquakes: Version 1.0, U.S. Geological Survey
in other regions of PI for which no records are available. Open-File Report 96-80-A.
Under such circumstances, the stochastic model of Boore Boore, D. M. (2003). Simulation of ground motion using the stochastic
(1983) to generate PGA values synthetically, incorporating method, Pure and Applied Geophysics160, 635-675.
uncertainties in the seismological parameters of the region, is Boore, D. M. and G. M. Atkinson (1987). Stochastic prediction of
ground motion in eastern North America, Bulletin of the Seismolog-
a viable alternative. The model can incorporate inhomogene-
ical Society afAmerica 4, 460--477.
ity in the quality factors and uncertainty in other model Boore, D. M. and J. Boatwright (1984). Average body-wave radiation
parameters. coefficients, Bulletin of the Seismological Society of America
Here PI is divided into three parts, namely the Koyna- 74,1,615-1,621.
Warna, western-central, and southern regions, for which Brune, J. (1970). Tectonic stress and the spectra of seismic shear waves
from earthquakes, Journal of Geophysical Research 75,
quality factors are available in the literature. Synthetic ground
4,997-5,009.
motion is generated separately for the three regions, consist- Campbell, K. W. (2003). Engineering models of strong ground motion,
ing of 10,100 samples covering the magnitude range of 4 to in W. E Chen, C. S. Scawthorn, and J. F. Arros (editors), Earth-
8, for bedrock (Vs = 3.6 km/s) site conditions. Regression quake Engineering Handbook, Boca Raton, FL: CRC Press,
analysis is carried out on the synthetic data to derive new 759-8O3.
Chandra, U. (1977). Earthquakes of peninsular India: A seismotectonic
regional attenuation relationships. An attenuation formula
study, Bulletin of the Seismological Society of America 67,
valid for PI in general has been proposed. Site-correction fac- 1,387-1,413.
tors for hard-rock (1.5 < 1/30 < 3.6 km/s), rock (0.76 < 1/30 < Cramer, C. H. and A. Kumar (2003). 2001 Bhuj, India earthquake
1.5 km/s) and firm-rock (V30 = 0.76 km/s) sites have been engineering seismoscope recordings and eastern North America
obtained. The available past records are insufficient to vali- ground-motion attenuation relations, Bulletin of the Seismological
date the new attenuation relation, since multiple recordings Society ofAmerica 93, 1,390-1,394.
Dahle, A., H. Bungum, and L. B. Kvamme (1990). Attenuation models
for the same event are lacking in PI. Hence to judge the atten- inferred from intraplate earthquake recordings, Earthquake Engi-
uation relation, PGA values estimated from SRR data of the neering and Structural Dynamics 19, 1,125-1,141.
Kutch earthquake of 26 January 2001 are compared with the Department of Earthquake Engineering (2001). Strong Motion Data
values estimated from the empirical attenuation relation. The from Kutch Earthquake of January 26,2001, Report, University of
Roorkee, June 2001, 1-27.
comparison is found to be favorable within statistical limits.
Frankel, A., C. Mueller, T. Barnhard, D. Perkins, E. Leyendecker, N.
The new attenuation equation is also compared with rela- Dickman, S. Hanson, and M. Hopper (1996). National Seismic-
tions derived for ENA. The present results indicate similari- Hazard Maps: DocumentationJune 1996, U.S. Geological Survey
ties in the attenuation between PI and ENA. The new Open-File Report 96-532, 110 pp.
attenuation relation proposed for PI needs to be revised as Guha, S. K. and P. C. Basu (1993). Catalogueof Earthquakes (M> 3.0)
and when more SMA data are obtained. B in PeninsularIndia, 1993, Atomic Energy Regulatory Board, Tech-
nical Document No. TD/CSE-1, 70 pp.
Guha, S. K., P. D. Gosari, M. M. Varma, S. D. Agarwal, J. G. Padale,
ACKNOWLEDGMENTS and S. C. Marwadi (1970). Recent Seismic Disturbances in the
Shivajisagar Lake Area of the Koyna HydroelectricProject, Maharash-
Thanks are due to Dr. C. H. Cramer for his useful comments tra, India, Report, Central Water and Power Research Station,
Poona-24, India.
and suggestions for improving the original version of the
Gupta, I. D., R. G. Joshi, and V. Rambabu (1992). Analysis of Some Sig-
paper. The work reported here has been supported by the nificant Accelerograms of Koyna Dam Earthquakes Using Improved
Department of Atomic Energy, Government of India under Data Processing Techniques, Technical Memorandum, Ministry of
grant No. 2002/36/5/BRNS. Water Resources, Central Water and Power Research Station.
Hwang, H. and J.-R. Huo (1997). Attenuation relations of ground
motion for rock and soil sites in eastern United States, Soil Dynam-
REFERENCES ics and Earthquake Engineering 16, 363-372.
Hwang, H., H. Lin, and J.-R. Huo (1997). Site coefficients for design
Atkinson, G. M. and D. M. Boore (1995). Ground-motion relations for of buildings in eastern United States, Soil Dynamics and Earth-
eastern North America, Bulletin of the SeismologicalSocietyofAmer- quake Engineering 16, 29-40.
ica 85, 17-30. Joyner, W. B. and D. M. Boore (1981). Peak horizontal acceleration and
BSSC (2001). NEHRP RecommendedProvisionsfor Seismic Regulations velocity from strong motion records including records from the
for New Buildings and Other Structures2000 Edition, Part 1: Provi- 1979, Imperial valley, California earthquake, Bulletin ofthe Seismo-
sions, prepared by the Building Seismic Safety Council for the Fed- logical Society ofAmerica 71, 2,011-2,038.
eral Emergency Management Agency (Report No FEMA 368), Joyner, W. B. and D. M. Boore (2000). Recent developments in earth-
Washington, DC. quake ground motion estimation, Proceedingsof the Sixth Interna-

Seismological ResearchLetters July/August2004 Volume75, Number4 539

 tional Conference on Seismic Zonation, November 12-15, 2000, Saragoni, G. R. and G. C. Hart (1974). Simulation of artificial earth-
Palm Springs, Oakland, CA: EERI. quakes, Earthquake Engineering and Structural Dynamics 2,
Kaila, K. L. and D. Sarkar (1978). Atlas of isoseismal maps of major 249-267.
earthquakes in India, GeophysicalResearch Bulletin 16, 233-267. Seeber, L., J. G. Armbruster, and K. H. Jacob (1999). Probabilistic
Kriniztsky, E. L., J. P. Gould, and E H. Edinger (1993). Fundamentals Assessment of Seismic Hazard for Maharashtra, Government of
of Earthquake-resistant Construction, New York: Wiley. Maharashtra, unpublished report.
Kumar, A., S. Basu, S. K. Thakkar, M. Shrikhande, P. Agarwal, J. Das, Singh, S. K., B. K. Bansal, S. N. Bhattacharya, J. F. Pacheco, R. S. Dat-
and D. K. Paul (2001). Strong motion records of Bhuj earthquake, tatrayam, M. Ordaz, G. Suresh, Kamal, and S. E. Hough (2003).
International Conference on Seismic Hazard with Particular Ref- Estimation of ground motion for Bhuj (26th January 2001,
erence to Bhuj Earthquake, New Delhi, October 3-5, 2001, Mw=7.6) and for future earthquakes in India, Bulletin of the Seis-
10pp. mological Society ofAmerica 93, 353-370.
Mandal, E and B. K. Rastogi (1998). A frequency-dependent relation of Singh, S. K., M. Ordaz, R. S. Dattatrayam, and H. K. Gupta (1999). A
coda Q, for Koyna-Warna region, India, Pure andApplied Geophys- spectral analysis of the 21 May 1997, Jabalpur, India, earthquake
ics 153, 163-177. (Mw = 5.8) and estimation of ground motion from future earth-
Narula, P. L., S. K. Acharyya, and J. Banerjee (2000). Seismotectonic quakes in the Indian shield region, Bulletin of the Seismological
Atlas of India and Its Environs, Geological Survey of India. Society ofAmerica 89, 1,620-1,630.
Rao, B. R. and E S. Rao (1984). Historical seismicity of peninsular Toro, G., N. Abrahamson, and J. Schneider (1997). Model of strong
India, Bulletin of the Seismological Society of America 74, ground motion in eastern and central North America: Best esti-
2,519-2,533. mates and uncertainties, SeismologicalResearch Letters 68, 41-57.
Rao, R., C. V. Seshamma, and E Mandal (1998). Estimation of Coda 0r
and Spectral Characteristics of Some Moderate Earthquakes of South- Department of Civil Engineering
ern Indian Peninsula, unpublished report.
Rastogi, B. K., R. K. Chadka, and C. S. P. Sarma (1995). Investigations
Indian Institute of Science
of June 7, 1988 earthquake of magnitude 4.5 near Idukki Dam in Bangalore 560012
southern India, Pure andApplied Geophysics 145, 109-119. India
rni @dviL iisc.ernet, in

540 SeismologicalResearchLetters Volume75, Number4 July/August2004