Professional Documents
Culture Documents
F
Jianghai Xia a,*, Richard D. Miller a, Choon B. Park a, Gang Tian b
O
2
a
3 Department of Geology, Kansas Geological Survey, The University of Kansas, 1930 Constant Ave.,
O
4 Campus West, Lawrence, KS 66047-3726, USA
b
5 Department of Geophysics, Jinlin University, PR China
6
7
PR
Received 19 December 2001; accepted 26 September 2002
8 Abstract
9 High-frequency (z 2 Hz) Rayleigh wave phase velocities can be inverted to shear (S)-wave velocities for a layered earth
10
11 D
model up to 30 m below the ground surface in many settings. Given S-wave velocity (VS), compressional (P)-wave velocity
(VP), and Rayleigh wave phase velocities, it is feasible to solve for P-wave quality factor QP and S-wave quality factor QS in a
TE
12 layered earth model by inverting Rayleigh wave attenuation coefficients. Model results demonstrate the plausibility of inverting
13 QS from Rayleigh wave attenuation coefficients. Contributions to the Rayleigh wave attenuation coefficients from QP cannot be
14 ignored when Vs/VP reaches 0.45, which is not uncommon in near-surface settings. It is possible to invert QP from Rayleigh
15 wave attenuation coefficients in some geological setting, a concept that differs from the common perception that Rayleigh wave
16 attenuation coefficients are always far less sensitive to QP than to QS. Sixty-channel surface wave data were acquired in an
EC
17 Arizona desert. For a 10-layer model with a thickness of over 20 m, the data were first inverted to obtain S-wave velocities by
18 the multichannel analysis of surface waves (MASW) method and then quality factors were determined by inverting attenuation
19 coefficients.
20 D 2002 Published by Elsevier Science B.V.
21
R
23
O
24
25 1. Introduction ogy. A desire to understand the attenuative properties 32
of the earth are based on the observations that seismic 33
C
26 The most common measure of seismic wave at- wave amplitudes are reduced as waves propagates 34
27 tenuation is the dimensionless quality factor Q and its through an elastic medium. This reduction is generally 35
inverse (dissipation factor) Q 1. The quality factor
N
31 as well as in oil exploration and earthquake seismol- saturation (Toksöz and Johnston, 1981). To fully 39
understand seismic wave propagation in the earth, 40
the quality factors are parameters that must be known. 41
* Corresponding author. Tel.: +1-785-864-2057; fax: +1-785-
High-frequency Rayleigh waves possesses informa- 42
864-5317. tion of the shear (S)-wave velocity (VS) and the 43
E-mail address: jxia@kgs.ku.edu (J. Xia). quality factors of near-surface materials. 44
F
analysis of surface waves (MASW), undertaken by 57
O
the Kansas Geological Survey (KGS) was designed to 58
estimate near-surface S-wave velocities from high- 59
frequency Rayleigh waves (Fig. 2): (1) acquisition 60
O
of multichannel high-frequency (z 2 Hz) broad band 61
Rayleigh waves; (2) creation of efficient and accurate 62
PR
algorithms organized in a basic data processing 63
sequence designed to extract Rayleigh wave disper- 64
sion curves from Rayleigh waves (Park et al., 1999a); 65
and (3) development of stable, robust, and efficient 66
inversion algorithms for inverting phase velocities of 67
Rayleigh waves for near-surface S-wave velocity 68
46 surface and are usually characterized by relatively low surface wave analysis increases the resolution (or 75
47 velocity, low frequency, and high amplitude energy accuracy) of the inverted S-wave velocities (Xia et 76
48 (Sheriff, 1991, p. 143). Estimates of the S-wave al., 2000b). After successfully determining a near- 77
R
R
O
C
N
U
Fig. 2. A diagram of the MASW method. Multichannel raw field data, which contain enhanced Rayleigh wave signals, are acquired. Rayleigh
wave phase velocities are extracted from the field data through a direct wavefield transformation method by Park et al. (1998). The phase
velocity, finally, is inverted for a shear-wave velocity profile (VS vs. depth).
ARTICLE IN PRESS
J. Xia et al. / Journal of Applied Geophysics 1398 (2002) 1–9 3
78 surface VS profile from Rayleigh waves, the feasibility factors of near-surface materials. The relationship 120
79 of calculating near-surface Q from high-frequency between Rayleigh wave attenuation coefficients and 121
80 Rayleigh wave attenuation coefficients can be ana- the quality factors for P- and S-waves of a layered 122
81 lyzed. model were given by Anderson et al. (1965) as: 123
82 Laboratory experiments (Johnston et al., 1979)
83 show that Q may be independent of frequency over pf
aR ð f Þ ¼
84 a broad bandwidth (10 2 – 107 Hz), especially for CR2 ð f Þ
85 some dry rocks. Q 1 in liquids, however, is propor- " #
F
X
n X
n
86 tional to frequency so that in some highly porous and Pi ð f ÞQ1
Pi þ Si ð f ÞQ1
Si ; ð2Þ
permeable rocks Q 1 may contain a frequency-
O
87 i¼1 i¼1
88 dependent component. This component may be negli-
89 gible at seismic frequency, even in unconsolidated where 125
124
O
90 marine sediments (Johnston et al., 1979). Mitchell
91 (1975) investigated Q structure of the upper crust in BCR ð f Þ
Pi ð f Þ ¼ VPi ; ð3Þ
BVPi
PR
92 the North America by inverting Rayleigh wave attenu-
93 ation coefficients in a layered earth model. In his 127
126
94 work, Q was independent of frequency. Although BCR ð f Þ
Si ð f Þ ¼ VSi ; ð4Þ
95 some authors suggest that near-surface Q may be BVSi
96 frequency dependent (Jeng et al., 1999), we will
97 follow the laboratory results (Johnston, 1981) and aR( f ) is Rayleigh wave attenuation coefficients in 128
129
98
99
Mitchell’s (1975) work that Q is independent of
frequency, allowing determination of Q as a function
D
1/length, and f is frequency in Hz. QPi and QSi are the
quality factors for P- and S-waves of the ith layer,
130
131
TE
100 of depth based on amplitude attenuation of Rayleigh respectively; VPi and VSi are the P-wave velocity and S- 132
101 wave data. In this paper, we will examine the relation- wave velocity of the ith layer, respectively; CR( f ) is 133
102 ship between Rayleigh wave attenuation coefficients Rayleigh wave phase velocity; and n is the number of 134
103 and compressional P-wave and S-wave quality factors layers of a layered earth model. 135
EC
104 ( QP and QS) through forward modeling. The model- We adopted Kudo and Shima’s (1970) work to 136
105 ing will be used to develop a quantitative description calculate the attenuation coefficients. The attenuation 137
106 of the contributions to Rayleigh wave attenuation coefficient is defined by 138
107 coefficients from QP and QS.
Aðx þ dxÞ ¼ AðxÞeadx ;
R
ð5Þ
h qffiffiffiffiffiffiffiffii
pf ðxþdx; f Þ xþdx
Q¼ ; ð1Þ ln WW ðx; f Þ x
N
av aR ð f Þ ¼ ; ð6Þ
dx
112
113 where v, f, and a are the velocity, the frequency, and the
U
114 attenuation coefficient of the plane wave, respectively. where aR( f ) is the Rayleigh wave attenuation coef- 145
146
115 To determine Q as a function of depth in near-surface ficient as a function of frequency f, W is the amplitude
qffiffiffiffiffiffiffiffiffiffi 147
116 materials (up to 30 m), the assumption of homogeneity of a specific frequency, and ðxþdxÞ
x is a scaling factor 148
117 is no longer valid because of complexity of the near- in calculating the attenuation coefficient. 149
118 surface geology. Utilization of high-frequency Ray- In the following section, contributions of QP and 150
119 leigh waves (z 2 Hz) is essential in finding the quality QS to attenuation coefficients of Rayleigh waves will 151
ARTICLE IN PRESS
4 J. Xia et al. / Journal of Applied Geophysics 1398 (2002) 1–9
F
157 Eqs. (3) and (4) represent the rate of change of
O
158 Rayleigh wave attenuation coefficients aR( f ) to dis-
159 sipation factors QP 1 and QS 1 of the ith layer,
160 respectively. Pi is the product of the P-wave velocity
O
161 of the ith layer and the partial derivative of Rayleigh
162 wave phase velocities with respect to P-wave velocity
PR
163 of the ith layer. Si is the product of the S-wave
164 velocity of the ith layer and the partial derivative of
165 Rayleigh wave phase velocities with respect to S-
166 wave velocity of the ith layer. Pi and Si totally control
167 the sensitivity of Rayleigh wave attenuation coeffi-
168 cients to QP 1 and QS 1.
169
170
A six-layer model (Xia et al., 1999) is employed to
analyze contributions to Rayleigh wave attenuation
D
TE
171 coefficients from QP and QS (Fig. 3). Letting VS
Fig. 4. QP contributions to Raleigh wave attenuation coefficients (a)
and QS contributions to Raleigh wave attenuation coefficients (b).
QP contributions become significant when VS/VP is about 0.5.
EC
F
directly to solve QP and/or QS from Rayleigh wave 230
O
attenuation coefficients. In many cases, only a single 231
iteration is necessary to obtain quality factors. Here 232
we discuss an algorithm from Menke (1984) including 233
O
our introduction of a damping factor. Our inversion 234
Fig. 5. Sensitivity of Rayleigh wave attenuation coefficients to QP problem can be described by the following system: 235
PR
and QS with VS/VP being equal to 0.5.
! !
A X ¼ B ðxi > 0Þ; ð7Þ
193 reduction in QP and/or QS (3.75, 7.5, 9.0, 11.25, 15.0,
!
194 and 18.75 from the top layer to the half space) results where X is an inverse of quality factors (a model 236
237
!
195 in the relationship shown in Fig. 5. With a 25% vector 1/Q) with xi as the ith component, B is 238
196 reduction in QP, the relative increases in Rayleigh attenuation coefficients (a data vector), and A is a 239
197
198
wave attenuation coefficients are in the range of 4–
20%, averaging 12% from 5 to 35 Hz. For the same (2).
D
data kernel matrix (Menke, 1984) determined by Eq. 240
241
TE
199 reductions in QS, the relative increases in Rayleigh Eq. (7) will provide accurate QP and QS if attenu- 242
200 wave attenuation coefficients are in the range of 9– ation coefficients contain no error as a synthetic 243
201 23% with an average of 17% from 5 to 35 Hz. The example shows in the following section. Solutions 244
202 overall relative increases in Rayleigh wave attenuation of Eq. (7) are not guaranteed to exist or solutions may 245
EC
203 coefficients due to a 25% reduction in both QP and QS possess an unacceptable error when attenuation coef- 246
204 are almost the same at 28% in relative change within ficients possess errors. Mitchell (1973, 1975) dis- 247
205 the frequency range of 5 –35 Hz. For a water saturated cussed the method of solving the inverse problem 248
206 sandstone, QP/QS may reach 2 (Johnston, 1981). In presented in Eq. (7). A damping factor k is introduced. 249
R
209 QS.
where I is the unit matrix. k is set to be a small value 250
251
(say 10 7) at the beginning of the inversion. Based on 252
O
210 4. Inversion system inverted results of QP and/or QS, k will be systemati- 253
cally increased until smooth solutions are obtained. 254
C
216 1999) and finding near-surface P-wave velocities by version system (Eq. 8) working properly. With a six- 257
217 other seismic methods, such as reflection (Hunter et layer model, this example assumes known P-wave and 258
218 al., 1984; Steeples and Miller, 1990), refraction S-wave velocities (Fig. 6a), error-free attenuation 259
219 (Palmer, 1980), and/or tomography methods (Zhang coefficients (labeled ‘‘Measured’’ in Fig. 6b), QS (5, 260
220 and Toksöz, 1998; Ivanov et al., 2000), the dissipation 10, 12, 15, 20, and 25 for layer one to the half space, 261
221 factors ( QP 1 and QS 1) can be inverted directly for respectively), and QP (twice of QS). Attenuation 262
ARTICLE IN PRESS
6 J. Xia et al. / Journal of Applied Geophysics 1398 (2002) 1–9
F
(Fig. 7). Geophones were deployed at 1.2-m intervals 271
O
with a nearest offset of 4.8 m. The seismic source was 272
an accelerated weight drop designed and built by the 273
KGS. A record length of 1024 ms at a 1-ms sample 274
O
interval was selected. 275
The number of layers of a model should be deter- 276
PR
mined by the quality of data (Xia et al., 2000b). A layer 277
in a model is usually not a geological layer. In most 278
case, we choose 10 to 15 layers in a model to obtain 279
certain resolution for an inverted model. For this 280
D
TE
EC
R
R
O
C
N
F
ratio of VS/VP for the model is approximately 0.4, 289
O
only QS can be confidently inverted from attenuation 290
coefficients. Under the assumption that QP was equal 291
to twice QS, we inverted attenuation coefficients to 292
O
obtain QS (Fig. 8c). Attenuation coefficients calcu- 293
lated from inverted quality factors QS were labeled 294
PR
‘‘Final’’ in Fig. 8b. Inverted QS results suggested 295
that there is a highly attenuating layer at a depth of 296
12.5 m. 297
326 in 6% error in S-wave velocity), Eq. (8) has less Johnston, D.H., 1981. Attenuation: a state-of-the-art summary. In: 372
327 stability. Hence, accurate calculation of Rayleigh Toksöz, M.N., Johnston, D.H. (Eds.), Seismic Wave Attenua- 373
tion, pp. 123 – 135. 374
328 wave attenuation coefficients is critical. On the other Johnston, D.H., Toksöz, M.N., 1981. Definitions and terminology. 375
329 hand, the inversion system (Eq. (8)) is more stable In: Toksöz, M.N., Johnston, D.H. (Eds.), Seismic Wave Attenu- 376
330 than AVO (amplitude versus offset) analysis studied ation, pp. 1 – 5. 377
331 and practiced in the oil industry for the last 20 years Johnston, D.H., Toksöz, M.N., Timur, A., 1979. Attenuation of 378
332 (Hilterman, 2001). Jin et al. (2000) concluded that in seismic waves in dry and saturated rocks: II. Mechanisms. Geo- 379
physics 44, 691 – 711. 380
333 AVO analysis, a 10% error in incident angles could
F
Kudo, K., Shima, E., 1970. Attenuation of shear wave in soil. Bull. 381
334 result in a 40% error in reflection coefficients. Because Earthq. Res. Inst. 48, 145 – 158. 382
O
335 our geophysical community accepts AVO practices, Menke, W., 1984. Geophysical Data Analysis—Discrete inversion 383
336 we should be more comfortable with quality factors Theory. Academic Press, New York. 384
337 that are inverted from Rayleigh wave attenuation Miller, R.D., Xia, J., Park, C.B., Ivanov, J., 1999. Multichannel 385
O
analysis of surface waves to map bedrock. Lead. Edge 18, 386
338 coefficients. 1392 – 1396. 387
Mitchell, B.J., 1973. Surface wave attenuation and crustal anelas- 388
PR
ticity in central North America. Bull. Seismol. Soc. Am. 63, 389
339 Acknowledgements 1057 – 1071. 390
Mitchell, B.J., 1975. Regional Rayleigh wave attenuation in North 391
America. J. Geophys. Res. 80, 4904 – 4916. 392
340 We thank Prof. J.A. Canas and one anonymous
Nazarian, S., Stokoe II, K.H., Hudson, W.R., 1983. Use of spectral 393
341 reviewer for their critical and constructive reviews, analysis of surface waves method for determination of moduli 394
342 which improved the paper. The authors thank Mary 395
343
344
Brohammer and Julia Shuklaper for their assistance in
manuscript preparation and submission.
D
and thicknesses of pavement systems. Transp. Res. Rec. 930,
38 – 45.
Palmer, D., 1980. The Generalized Reciprocal Method of Seismic
396
397
TE
Refraction Interpretation. Society of Exploration Geophysicists, 398
Tulsa, OK. 399
Park, C.B., Miller, R.D., Xia, J., 1996. Multi-channel analysis of 400
345 References surface waves using Vibroseis (MASWV). Expanded Abstracts 401
of Technical Program with Biographies, Society of Exploration 402
EC
346 Aki, K., Richards, P.G., 1980. Quantitative Seismology—Theory Geophysicists, 66th Annual Meeting, Denver, CO, pp. 68 – 71. 403
347 and Methods. Freeman, San Francisco. Park, C.B., Miller, R.D., Xia, J., 1998. Imaging dispersion curves of 404
348 Anderson, D.L., Ben-Menahem, A., Archambeau, C.B., 1965. At- surface waves on multi-channel record. Technical Program with 405
349 tenuation of seismic energy in upper mantle. J. Geophys. Res. Biographies, Society of Exploration Geophysicists, 68th Annual 406
350 70, 1441 – 1448. Meeting, New Orleans, LA, pp. 1377 – 1380. 407
R
351 Dorman, J., Ewing, M., 1962. Numerical inversion of seismic sur- Park, C.B., Miller, R.D., Xia, J., 1999a. Multichannel analysis of 408
352 face wave dispersion data and crust-mantle structure in the New surface waves. Geophysics 64, 800 – 808. 409
353 York – Pennsylvania area. J. Geophys. Res. 67, 5227 – 5241. Park, C.B., Miller, R.D., Xia, J., Hunter, J.A., Harris, J.B., 1999b. 410
R
354 Hilterman, F.J., 2001. Seismic amplitude interpretation. Distin- Higher mode observation by the MASW method. Technical 411
355 guished Short Course 4. Society of Exploration Geophysicists Program with Biographies, Society of Exploration Geophysi- 412
356 and European Association of Geoscientists and Engineers. cists, 69th Annual Meeting, Houston, TX, pp. 524 – 527. 413
O
357 Hunter, J.A., Pullan, S.E., Burns, R.A., Gagne, R.M., Good, R.L., Sheriff, R.E., 1991. Encyclopedic Dictionary of Exploration Geo- 414
358 1984. Shallow seismic reflection mapping of the overburden- physics, 3rd ed. Society of Exploration Geophysicists, Tulsa, 415
359 bedrock interface with the engineering seismograph—some sim- OK. 416
C
360 ple techniques. Geophysics 49, 1381 – 1385. Steeples, D.W., Miller, R.D., 1990. Seismic-reflection methods ap- 417
361 Ivanov, J., Park, C.B., Miller, R.D., Xia, J., Hunter, J., Good, R.L., plied to engineering, environmental, and ground-water prob- 418
N
362 Burns, R.A., 2000. Joint analysis of surface-wave and refraction lems. In: Ward, S.H. (Ed.), Geotechnical and Environmental 419
363 events from river-bottom sediments. Technical Program with Geophysics 1. Society of Exploration Geophysicists, pp. 1 – 30. 420
364 Biographies, Society of Exploration Geophysicists, 70th Annual Stokoe II, K.H., Nazarian, S., 1983. Effectiveness of ground im- 421
U
365 Meeting, Calgary, Canada, pp. 1307 – 1310. provement from spectral analysis of surface waves. Proceeding 422
366 Jeng, Y., Tsai, J., Chen, S., 1999. An improved method of determin- of the Eighth European Conference on Soil Mechanics and 423
367 ing near-surface Q. Geophysics 64, 1608 – 1617. Foundation Engineering, Helsinki, Finland. 424
368 Jin, S., Cambois, G., Vuilermoz, C., 2000. Shear-wave velocity Toksöz, M.N., Johnston, D.H., 1981. Preface. In: Toksöz, M.N., 425
369 and density estimation from PS-wave AVO analysis: applica- Johnston, D.H. (Eds.), Seismic Wave Attenuation. 426
370 tion to an OBS dataset from the North Sea. Geophysics 65, Xia, J., Miller, R.D., Park, C.B., 1997. Estimation of shear wave 427
371 1446 – 1454. velocity in a compressible Gibson half-space by inverting Ray- 428
ARTICLE IN PRESS
J. Xia et al. / Journal of Applied Geophysics 1398 (2002) 1–9 9
429 leigh wave phase velocity. Expanded Abstracts of Technical Xia, J., Miller, R.D., Park, C.B., 2000b. Advantage of calculat- 445
430 Program with Biographies, Society of Exploration Geophysi- ing shear-wave velocity from surface waves with higher 446
431 cists, 67th Annual Meeting, Dallas, TX, pp. 1927 – 1930. modes. Technical Program with Biographies, Society of Ex- 447
432 Xia, J., Miller, R.D., Park, C.B., 1998. Construction of vertical ploration Geophysicists, 70th Annual Meeting, Calgary, Can- 448
433 section of near-surface shear-wave velocity from ground roll. ada, pp. 1295 – 1298. 449
434 Technical Program, The Society of Exploration Geophysicists Xia, J., Miller, R.D., Park, C.B., Hunter, J.A., Harris, J.B., Ivannov, 450
435 and The Chinese Petroleum Society, Beijing 98’ International J., 2002a. Comparing shear-wave velocity profiles from multi- 451
436 Conference, pp. 29 – 33. channel analysis of surface wave with borehole measurements. 452
437 Xia, J., Miller, R.D., Park, C.B., 1999. Estimation of near-surface Soil Dyn. Earthqu. Eng. 22 (3), 181 – 190. 453
F
438 shear-wave velocity by inversion of Rayleigh wave. Geophysics Xia, J., Miller, R.D., Park, C.B., Wightman, E., Nigbor, R., 2002b. 454
439 64, 691 – 700. A pitfall in shallow shear-wave refraction surveying. J. Appl. 455
O
440 Xia, J., Miller, R.D., Park, C.B., Hunter, J.A., Harris, J.B., 2000a. Geophys. 51 (1), 1 – 9. 456
441 Comparing shear-wave velocity profiles from MASW with Zhang, J., Toksöz, M.N., 1998. Nonlinear refraction traveltime to- 457
442 borehole measurements in unconsolidated sediments, Fraser mography. Geophysics 63, 1726 – 1737. 458
O
443 River Delta, B.C., Canada. J. Environ. Eng. Geophys. 5 (3),
444 1 – 13.
PR
D
TE
EC
R
R
O
C
N
U
459