Professional Documents
Culture Documents
Attenuation Low Freq Waves New Madrid
Attenuation Low Freq Waves New Madrid
BY OTTO W. NUTTLI
ABSTRACT
From a time-domain study, the amplitude of lO-Hz P waves in the New Madrid
seismic zone was found to fall off as the inverse 1.4 power of the epicentral
distance. The amplitude of lO-Hz Lg waves was found to decay as for an Airy
phase with a coefficient of anelastic attenuation of 0.006 k m - ' . In almost all
cases the Lg motion was found to be larger than that of P, even at epicentral
distances of 5 km and less. Thus design earthquake motions need to be
concerned with Lg waves, whose coda also have a longer duration than those of
the P waves.
The apparent Q for the l O-Hz Lg waves'is 1500, identical to that found
previously by Nuttli (1973) for higher mode 1-Hz Lg waves. The attenuation of
lO-Hz Lg waves is sufficiently small that one must pay attention to their damage
potential at distances as large as a few hundred kilometers.
INTRODUCTION
Damaging ground accelerations usually are associated with relatively high-fre-
quency wave motion, in the range of I to 10 Hz. A knowledgeof the attenuation of
this wave motion is required for the specificationof the ground motion at a given
site, for the purpose of designingstructures.
Previously, Nuttli (1973) estimated the anelastic attenuation coefficientof l-Hz
Lg wavesfor the central United States to be 0.07 deg-I (0.0006 kin-l). He determined
that at frequenciesnear I Hz the Lg wave carried the maximumamplitude, at least
for distances of 25 km and greater. Data for this study came from WWSSN stations
or eastern North America stations which had short-period seismographs with
response similar to those of the WWSSN stations. Since the summer of 1974 a
microearthquake network has been operating in the New Madrid seismic zone
(Stauder et al., 1976). As the peak response of the microearthquake seismographs
occurs at frequenciesnear I0 Hz, the network provides ideal data for a time-domain
study of the attenuation of 10-Hz waves. This paper is concerned with an analysis
of the P- and Lg-waveattenuation of these high-frequencywaves.
THE DATA
Table 1 lists the hypocentral coordinates of the 77 earthquakes whose data were
used. In general, the earthquakes studied are associated with the New Madrid
faulted zone, which is about 175 km long and 25 to 50 km wide. Fourteen stations
recording on Develocorder film, with film reader gains of 8.5 × 104 to 1.4 × 106 at 10
Hz, provided the basic amplitude data. Epicentral distances varied from approxi-
mately 3 to 250 kin, although for any given earthquake the range of distances usually
was less. Only the vertical component of ground motion was recorded and measured.
Amplitude measurements were made directly from the image of the Develocorder
film as seen on the screen of the film reader. For the P-wave portion of the
seismogram the largest amplitude was measured. Occasionally the P-wave motion
consisted of a single spike, but more commonly it was composed of a train of waves
persisting in amplitude until the arrival of the S or Lg waves. To be consistent with
343
344 OTTO W. N U T T L I
TABLE 1
LIST OF EARTHQUAKESSTUDIED
t h e a m p l i t u d e s t u d y o f 1-Hz w a v e s ( N u t t l i , 1973), t h e Lg a m p l i t u d e m e a s u r e d w a s
t h a t o f t h e m a x i m u m s u s t a i n e d m o t i o n , i.e., t h e l a r g e s t a m p l i t u d e e q u a l e d o r
e x c e e d e d b y a t l e a s t t h r e e c y c l e s of w a v e m o t i o n . T h e r e a s o n for u s i n g m a x i m u m
s u s t a i n e d r a t h e r t h a n p e a k m o t i o n is t h a t t h e m a x i m u m s u s t a i n e d m o t i o n is
c o n s i d e r e d a b e t t e r m e a s u r e of t h e size of t h e e a r t h q u a k e a n d of t h e d a m a g e
p o t e n t i a l o f t h e L g w a v e s t h a n t h e single p e a k a m p l i t u d e , b e c a u s e t h e single p e a k
is o f s h o r t d u r a t i o n a n d m a y r e s u l t f r o m f o r t u i t o u s c o n s t r u c t i v e i n t e r f e r e n c e .
A ~ A- g . (1)
I , I ' i'' 1 , t , I
IO0
~\\ '\
\\'~\\x
E Io \\\~\\
x \\\~\\
\\~ \\
, | I i
I0 I00 000
A (KM)
FIG. 1. Maximum P-wave amplitudes versus epicentral distance, for earthquake no. 63. The solid-line
curve of slope -2.4 is fitted to the data points indicated by x's. The dashed lines are separated in the
ordinate direction by one standard deviation from the solid line.
standard deviations, for b o t h the nonAiry (left) and Airy (right) cases. Although in
each case the median lies in the same range (0.201 to 0.250 logarithmic units), the
standard deviations are somewhat smaller for the Airy case. This can also be seen
from an inspection of Table 2. T h u s we conclude t h a t the 10-Hz Lg-wave data are
better fitted by an Airy phase relationship.
T h e amplitude deviations from the P and Lg curves can also be used to estimate
station corrections and personal bias in the evaluation of N and y. T o do this, the
sum of the deviations, taking regard of their sign, was determined for each station
and for specific ranges of epicentral distance. Tables 3, 4, and 5 give the results.
Table 3 presents the average deviation of the P-amplitude data from the straight-
line curves for each station. A value of 125 per cent indicates t h a t on the average
the observed amplitude at t h a t station exceeded the value expected from the
straight-line curve by 25 per cent. F r o m the table it can be seen t h a t stations DWM,
ECD, ELC, and P G A had lower t h a n expected amplitudes, by as m u c h as a factor
of 2, and t h a t stations OKG, POW, and R M B had higher t h a n expected amplitudes,
by as m u c h as a factor of 1½.
Table 3 also presents the average deviations of the P-amplitude data from the
A T I M E - D O M A I N S T U D Y OF T H E A T T E N U A T I O N OF 10-HZ W A V E S 347
TABLE 2
ATTENUATION OF P AND Lg WAVES
P Waves Lg Waves
Earthquake No.
;,(nonAiry) ~dAiry)
a N S.D. log A n S.D. log A S.D. log A
(kin-') (kin-~)
1 3 1,7 0,367 3 0.002 0.119 0.006 0.108
2 5 2.1 0.069 4 -- -- 0.008 0.213
3 6 1.9 0.284 5 0.008 0.153 0.008 0.152
4 6 2.0 0.231 7 0,008 0,181 0.008 0.181
5 9 1.5 0.295 9 0.002 0.169 0.006 0.158
6 4 0.8 0.106 5 -- -- 0.008 0.443
7 8 2.1 0.383 8 0.008 0.283 0.008 0.262
8 11 1.6 0.407 6 0.008 0.257 0.008 0.250
9 6 1.9 0.518 5 0.008 0.095 0.010 0.118
10 2 0.9 0.000 2 0,008 0.000 0.010 0.000
11 5 1.8 0.109 7 0.008 0.214 0.008 0.187
12 8 1.2 0.110 10 0.008 0.226 0,008 0.220
13 11 2.1 0.246 11 0.008 0.188 0.008 0,181
14 9 1.8 0.306 11 0,008 0.176 0.008 0.201
15 10 2.1 0.150 10 0.006 0.171 0.008 0.161
16 9 2.1 0.242 9 0.008 0.157 0.008 0.149
17 10 1.0 0.443 10 -- -- 0.008 0.375
18 11 1.9 0.268 . . . .
19 7 1,7 0.232 7 0.008 0,215 0.008 0.200
20 9 1.6 0.179 9 0.002 0.251 0.008 0.274
21 11 1.7 0.228 . . . .
22 10 2.4 0.443 10 0.010 0.363 0.008 0.286
23 10 1.2 0.302 10 0.010 0.280 0.908 0.243
24 7 2.7 0.475 7 -- -- 0.008 0.224
25 9 1.2 0.263 9 0.008 0.233 0.006 0.197
26 6 2.2 0.502 5 0.008 0.151 0.008 0.127
27 10 1.7 0.292 10 0.006 0.203 0.006 0.185
28 8 2.0 0.606 8 0.010 0.310 0.008 0.267
29 6 2.0 0.495 7 0.008 0.241 0.008 0.219
30 6 2.8 0.625 7 0.008 0.428 0.008 0.385
31 8 1.9 0.410 8 0.008 0.193 0,008 0.180
32 9 1,3 0.349 9 0.010 0.348 0,008 0.311
33 9 2.1 0,375 9 0.002 0.248 0.008 0.283
34 8 2.2 0.408 8 0.008 0.259 0.010 0.249
35 8 2.1 0.361 9 0.006 0.268 0.008 0.266
36 10 2.6 0.638 10 0.010 0.285 0.008 0.240
37 13 1.5 0.122 13 0.008 0.194 0.008 0.166
38 12 1.9 0.366 12 0.006 0.237 0.008 0.239
39 10 2.2 0.364 10 0.006 0.266 0.008 0.277
40 11 2.2 0.360 11 0.020 0.313 0.008 0.201
41 11 1.9 0.274 I1 0.096 0.243 0.008 0.251
42 8 1.7 0.281 . . . .
43 8 2.1 0.407 8 0.010 0.221 0.008 0,223
44 7 1.7 0.291 7 0.006 0.153 0.010 0.154
45 14 1.9 0.624 6 0.002 0.273 0.008 0.290
46 11 1.3 0.378 11 0.010 0.373 0.008 0.323
47 9 1.3 0.357 9 0.008 0.418 0.006 0.349
48 7 1,5 0.292 8 0.008 0.243 0.008 0.223
49 8 2.1 0.333 8 0.010 0,183 0.008 0.140
50 9 1.8 0.256 9 0.008 0.276 0.008 0.265
51 7 2.5 0.377 7 0.010 0.266 0.008 0.241
52 8 1.7 0.576 8 0.010 0.282 0.008 0.247
348 O T T O W. N U T T L I
T A B L E 2--continued
P Waves Lg Waves
Ea~hquakeNo,
y(nonAiry) ~,(Airy)
n N S.D. log A n (km_~) S.D. log A (km ') S.D. log A
~4
z
b-
86
C ! , L I I I l I i i i i i i i i i . i
1.0 1.5 2.0 2. 5
N
FIG, 2. Histogram of N values for the P-wave curves. The median N value is 1.8.
straight line curves for 25-km distance intervals. At small distances the log amplitude
deviations are negative, indicating lower than expected amplitudes. There is a
progression to more positive values as the distance increases. From this we can
conclude that the slope of the attenuation curve, corresponding to N -- 1.8, is too
large. Figure 6 shows the deviations, plotted as crosses, from the straight-fine curve
of slope -1.8. The dashed-line curve drawn through the crosses has a slope of -1.4,
indicating that a value of N = 1.4 in e~luation (1) will provide the best fit to the P-
wave amplitude data.
Tables 4 and 5 present the average deviation of the Lg-wave amplitude data from
the nonAiry and Airy curves, respectively, for y = 0.008 km -1. From both tables it
can be seen that DWM, ECD, ELC, and LST have significantly lower than expected
Lg amplitudes, and that CRU, GRT, OKG, POW, and RMB have significantly
higher than expected amplitudes. There is no obvious correlation between site
A TIME-DOMAIN STUDY OF THE A T T E N U A T I O N OF 10-HZ WAVES 349
, I , ,I,~ I r f I I Ill I ,
"x x"--,,
I0(
E
"""~\,~,,
"-,, \ "~,,
\\~\\ x
Io
\\x~\\
x \\\\
I ,I I I tJlll I I I I,,,,I ,
I0 I00
A (KM)
FIG. 3. Sustained maximum Lg-wave amplitudes versus epicentral distance, for earthquake no. 55.
The solid-line curve is a theoretical Airy phase attenuation curve with 7 - 0.008 km-L The dashed lines
are separated in the ordinate direction by one standard deviation from the solid line.
AIRY
60
oo
Z
9 40
>
n- NON-AIRY
bJ
co
@0
0
h
0
o20
z
(KM")
FIG. 4. Histograms of 7 values for the Lg-wave curves; left, nonAiry; r~ht, Airy. In both casesthe
median value of 7 is 0.008 kin-'.
AIRY
NON-AIRY
20
¢.0
z
o
to
\
m
o
LL
o
~5
E
0
I I
.100 2.100
i .300l ~,400 I---I
0 I00 200 300
I
.400
STANDARD D E V I A T I O N S
FIG. 5. Histograms of standard deviations of logarithm of Lg amplitudes; left, nonAiry; right, Airy. In
both cases the median value lies between 0.200 and 0.250 units.
TABLE 3
AVERAGE P-WAVE DEVIATIONS
AverageDeviation
Station
Logamplitudeunits Percentage
CRU +0.018 104.2
DON -0,008 98.2
DWM -0.257 55.4
ECD -0.254 55.7
ELC -0.281 52.3
GRT -0.010 97.7
LST -0,036 92.1
NKT -0.148 71.2
OKG +0.192 155,7
PGA -0.219 60,5
POW +0.107 127,9
RMB +0.198 157,8
WCK -0,057 87,6
DistanceInterval(kin)
0-25 -0,252 56.0
25.1-50 -0.235 58,2
50.1-75 --0.209 61,8
75.1-100 +0.050 112,1
100.1-125 +0.128 134.2
125.1-150 +0.099 125.7
150.1-175 +0.188 154.1
175.1-200 +0.174 149.3
>200 +0.140 138.1
distances are less t h a n expected and at the larger distances greater t h a n expected,
with respect to theoretical curves for y = 0.008 k m -1. Figure 7 shows the deviations
with respect to the Airy-phase curve. F r o m it we can conclude t h a t the dashed-line
curve, for which 7 = 0.006 k m -1, provides a b e t t e r fit to the data.
Previously, Nuttli (1973) h a d found t h a t y = 0.0006 k m -~ for 1-Hz Lg waves. T h e
a p p a r e n t quality factor (~ is given b y (~ = ~rf/Uy, where U is the group velocity and
f i s the frequency. Using 7 = 0.0006 k m -1 for 1-Hz and 7 -- 0.006 k m -1 for 10-Hz Lg
A T I M E - D O M A I N S T U D Y OF T H E A T T E N U A T I O N OF 10-HZ WAVES 351
, t v , ~ r i , i f I i ,,,| i , ~ f'
\
\X
X
A
co I 0 0
I--
.,..,.
)-
B=:
p-
.....
El
E:
'--" I0
-2
I0 I00
h (KM)
FIG. 6. Determination of slope correction for P-wave amplitudes. The solid-line curve has the median
slope, found by fitting the amplitude data by eye. The X's are the average deviations, for 25 km intervals,
from the solid-line curve. The dashed-line curve, which is the fit to the data represented by the x's, has
a slope o f - 1 . 4 .
TABLE 4
AVERAGE Lg-WAVE (NONAIRY) DEVIATIONS
Average Deviation
Station
Log amplitude units Percentage
CRU +0.071 117.8
DON +0.019 104.5
DWM -0.137 72,9
ECD -0.252 55.9
ELC -0.232 58.6
GRT +0.062 115.5
LST -0,148 71.2
NKT -0,059 87.2
OKG +0.252 178.6
PGA -0.049 89.4
POW +0,116 130.7
RMB +0.078 119.5
WCK --0.012 97.3
D i s t a n c e I n t e r v a l (kin)
TABLE 5
AVERAGE Lg-WAvE (AIRY) DEVIATIONS
AverageDeviation
Station
Log amplitudeunits Percentage
i i I 'I'I I , , I I I ill I I I I
'~
I-10o
ii:
I---
n~
,¢
10
~ = . 0 0 6
X • .008
waves, and U = 3.5 km/sec for both frequency waves, we get Q = 1,500 for both 1-
and 10-Hz waves.
Espinosa (1977), from a spectral analysis of the 1971 San Fernando earthquake
accelerograms, fitted straight-line curves to the logarithm of the spectral velocity
versus the logarithm of the epicentral distance. His curves for T -- 0.1 sec and 1 sec,
for the vertical component of the spectral velocity at distances of 10 to 100 km, have
slopes which are in substantial agreement with the values found for the New Madrid
zone, i.e., his slopes correspond to 7 = 0.0006 km -1 for 1-sec waves and 0.006 km -1
for 0.1-sec waves. Espinosa also gave a curve for 0.3-sec waves, which corresponds
to 7 = 0.020 km -1, implying a low ~) value for 0.3-sec period waves. It will be
interesting to see if spectral analyses of the New Madrid microearthquake data also
TABLE 6
A(P)/A(Lg)
Kilometers
indicate a relatively high anelastic attenuation for 0.3-sec period waves or if, rather,
they indicate a constant Q between periods of 0.1 and 1 sec.
For ground-motion predictions it is important to know which of the wave types
carries the largest amplitude at a given epicentral distance. At large distances it is
observed that the amplitude of the Lg-wave motion always exceeds that of the P-
wave motion. Table 6 presents the amplitude ratio A(P )/A(Lg) for all observations
at ~ < 35 km. From the table it can be seen that of the 94 observations in this
distance range only 18 correspond to the ratio exceeding unity. Of these, 10 could be
explained by an anomalously small Lg amplitude. Of the remaining 8, the largest
value of the ratio is 1.66. Thus, in general, we can say that over the entire range of
epicentral distance the Lg (or perhaps Sg) wave has the largest amplitude. Therefore
predictions of the maximum ground motion can be based upon the behavior of the
Lg wave, such as described by equation (2).
354 OTTO W. NUTTLI
A = Ao h-l4
where 5 is in kilometers, for the distance range of 3 to 250 km. The data were
obtained from microearthquakes of m~ approximately 1 to 2 as recorded on Devel-
ocorder film. For these small magnitude earthquakes the source region is of very
small dimensions, on the order of tens of meters. For larger earthquakes with
extended source regions the attenuation relations given above may break down for
small epicentral distances.
For almost all the observation points the amplitude of the Lg-wave motion
exceeded that of the P-wave motion, even at small epicentral distances. Thus for
predicting maximum ground motion one need only be concerned with the effects of
the Lg wave. Although duration of ground motion was not studied quantitatively in
this investigation, it was observed that the Lg-wave motion and its coda are of
longer duration than the P-wave motion, making the Lg motion comparatively even
more significant.
The high apparent Q for 10-Hz Lg waves, 1,500, is identical to that found
previously for 1-Hz Lg waves (Nuttli, 1973). By contrast, the Q of fundamental-
mode surface waves of the same frequency is about one order of magnitude lower,
resulting in larger absorption and more rapid attenuation with distance. Compared
to the higher-mode waves the high-frequency fundamental-mode waves are most
strongly excited by very shallow earthquakes (Herrmann and Nutfli, 1975). The
Illinois earthquake of August 14, 1965, with a focal depth of 1 km, provides an
example of such a shallow earthquake with high epicentral intensity and rapid
attenuation of intensity with epicentral distance (Nuttli and Zollweg, 1974).
Both the 1- and 10-Hz Lg waves are higher mode waves. Their high Q value is
typical of that found for the lower half of the crust in the central United States by
Herrmann and Mitchell (1975).
Aki and Chouet {1975), from a study of coda waves, concluded that Q at 10 to 20
Hz was of the order of 1,000 for both the regions of Tsukuba, Japan and Stone
Canyon, California. They attributed this high Q value to scattering of body waves
in the lithosphere. At 1 Hz they obtained a Q of approximately 200 at Tsukuba and
70 at Stone Canyon. They concluded that these lower Q values are indicative of the
scattering of surface waves in the upper crustal layers. Perhaps their data also could
be explained by scattering of higher-mode surface waves at the high frequencies and
of fundamental-mode surface waves at frequencies near 1 Hz. Herrmann (1975)
presented a time-domain coda model, based on empirical observations of the shape
of the coda envelope with time, which did not take account of the variation of the
coda envelope with frequency and epicentral distance. In a later paper (Herrmann,
1977) he examined the dependence of the coda frequency content upon the corner
frequency of the source spectrum, and obtained good agreement between observed
A TIME-DOMAIN STUDY OF T H E ATTENUATION OF 10-HZ WAVES 355
coda and calculated ones. His results support the conclusion that Q has a high value
for frequencies in the range of 1 to 10 Hz.
Although the details of the relation of Q and of the coefficient of anelastic
attenuation in the frequency band of 1 to 10 Hz remain to be resolved, the present
study and the others cited indicate that waves of 10-Hz frequency can produce
significant ground motions at distances of as much as a few hundred kilometers.
Their absorption is not nearly as great as the author thought it would be before this
study was carried out.
ACKNOWLEDGMENTS
I wish to thank R. B. Herrmann and B. J. Mitchell for helpful discussions on the nature of the Q
behavior in the crust. This research was supported in part by the Waterways Experiment Station of the
Corps of Engineers (Contract DACW39-77-C-0038), the Earth Sciences Section of the National Science
Foundation (Grant DES74-22852 A01) and the Division of Advanced Environmental Research and
Technology, National Science Foundation (Grant ENV76-20875). The data came from the New Madrid
Seismic Network, which is operated under U.S. Geological Survey contract 14-08-0001-15867.
REFERENCES
Aki, K. and B. Chouet (1975). Origin of coda waves: source, attenuation, and scattering effects, J.
Geophys. Res. 80, 3322-3342.
Espinosa, A. F. (1977). Particle-velocity attenuation relationship: San Fernando earthquake of February
9, 1971, Bull. Seism. Soc. Am. 67, 1195-1214.
Ewing, M., W. S. Jardetzky, and F. Press (1957). Elastic Waves in Layered Media, p. 358, McGraw-Hill,
New York.
Herrmann, R. B. (1975). The use of duration as a measure of seismic moment and magnitude, Bull.
Seism. Soc. Am. 65, 899-913.
Herrmann, R. B. (1977). A method for synthesis of the seismic coda of local earthquakes, Geophys. J. 43,
341-350.
Herrmann, R. B. and B. J. Mitchell (1975). Statistical analysis and interpretation of surface-wave
anelastic attenuation for the stable interior of North America, Bull. Seism. Soc. Am. 65, 1115-1128.
Herrmann, R. B. and O. W. Nuttli (1975). Ground-motion modelling at regional distances for earthquakes
in a continental interior, II. Effect of focal depth, azimuth and attenuation, Earthquake Eng. Struct.
Dyn. 4, 59-72.
Nuttli, O. W. (1973). Seismic wave attenuation and magnitude relations for eastern North America, J.
Geophys. Res. 78, 876-885.
Nuttli, O. W. and J. E. Zollweg (1974). The relation between felt area and magnitude for central United
States earthquakes, Bull. Seism. Soc. Am. 64, 73-85.
Stauder, W., M. Kramer, G. Fischer, S. Schaefer, and S. T. Morrissey (1976). Seismic characteristics of
southeast Missouri as indicated by a regional telemetered microearthquake array, Bull. Seism. Soc.
Am. 66, 1953-1964.