You are on page 1of 9


Construction and Building Materials xx (2002) xxxxxx


3 Assessment of the dynamic properties of highly saturated concrete using

one-sided acoustic tomography. Application in the Marathon Dam

6 Vassilios K. Karastathisa,*, Petros N. Karmisb, George Drakatosa, George Stavrakakisa
8 Institute of Geodynamics, National Observatory of Athens, P.O. Box. 20048, 118 10 Athens, Greece
9 Institute of Geology and Mineral Exploration, Mesogeion 70, 115 27 Athens, Greece

11 Received 10 September 2001; received in revised form 28 March 2002; accepted 17 April 2002


21 Abstract
Within the frames of a geophysical project aimed at assessing the Marathon Dam and its surrounding area, it was decided to
implement the one-sided acoustic refraction tomography method in the main interior tunnel of this concrete dam, in order to get
a picture about the dynamic properties of the concrete. This method was chosen since the implementation of the acoustic crosshole
tomography method with the sources and the receivers at different sides of the dam, was not feasible due to the existence of an
ornamental marble pavement. This marble pavement could not accept any physical destruction, but additionally it was of higher
28 seismic velocity than the interior material, resulting in problems on the correct application of this method. 2002 Elsevier
29 Science Ltd. All rights reserved.

31 Keywords: Non-destructive testing; Concrete; Dams; Tomography

34 55
1. Introduction general evaluation of its mechanical characteristics. The
method, in combination with the electrical tomography, 56

35 The monitoring of old large construction works such aimed to detect potential points that may enable water 57
36 as dams, bridges etc., has been intensified in recent infiltration within the structure. The acoustic and electric 58
37 years worldwide. Many of these works have long ago tomography methods, having been developed and estab- 59
passed their scheduled operational lifetime and the need lished in the field of Applied Geophysics, have been

38 60
39 for quality control becomes imperative, especially when recently introduced in the area of Non-Destructive Test- 61
40 these constructions have suffered from strong seismic ing investigations (NDT) for dams and other large 62
41 shocks. construction works w14x. 63
42 The Marathon dam, having completed 70 years of its The traditional destructive investigation methods 64

43 lifetime, was tested by the strong earthquake of Athens included the drilling of two to four boreholes, sited at 65
44 in September of 1999. A geophysical survey program the top of the dam and reaching its basement, in order 66
45 was conducted afterwards, aiming to the detection of to extract large diameter core samples (250300 mm) 67
46 leakages and the investigation of the geological condi- for laboratory testing. These samples were destructively 68
47 tions of its basement and surrounding area, and during tested, to determine their strength and elasticity module. 69
48 its course it was decided to extend the work into the The problem with the destructive methods is that despite 70
49 body of the dam, particularly at its main tunnel, using their high cost, they examine only a small fraction of 71
50 the methods of acoustic and electrical tomography. In the volume, usually not larger than 0.1% w3x. Addition- 72
51 view of the fact that very little information was availa- ally, these techniques are not capable of determining 73
52 ble, regarding the dynamic properties and strength char- anomalies to the construction of the dam. 74
53 acteristics of the concrete, it was agreed to conduct
The acoustic tomography besides the evaluation, by 75
54 surveys within the body of the dam, in order to have a
non-destructive means, of the strength and elasticity 76

493 *Corresponding author. Fax: q30-10-3490179. modulus, can also describe the characteristics of the 77
494 E-mail address: (V.K. Karastathis). internal structure of the construction material of the dam 78

0950-0618/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 5 0 - 0 6 1 8 0 2 . 0 0 0 2 1 - 1
2 V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx
79 133
and to determine its significant defects. In combination cells. The number and shape of the trapezoid cells are
80 with electrical tomography, it can also detect points of defined by the user. In this procedure, velocity values 134
81 leakage in the dam. are assigned to the cell corners. The velocity values 135
within the trapezoids come from a linear interpolation 136
82 2. The method between the four defined corner velocities. The Rayinvr 137
has been mostly used in crustal studies such as those 138
83 The mapping of the acoustic wave velocity within a reported in refs. w810x, but recently also in surface 139
84 medium, at various sections known as acoustic travel- scale problems w11x. 140
85 time tomography (or sonic tomography), may give In non-destructive testing, engineers basically use 14

86 important information with regard to the structure and acoustic tomography for scanning walls w1214x. In 142

87 condition of the medium. The velocity determination is most applications the sources and the receivers have 143
88 accomplished by measuring the time intervals taken by been set on opposite sides of the investigated body in 144
89 acoustic waves to travel from various sources to receiv- order to succeed the best ray coverage. When both 145
90 ers placed on the surface of the medium (geophones or sources and receivers are at the same side of a uniform 146
91 accelerometers). An acoustic ray is defined by each pair body, the investigation depth of the acoustic tomography 147

92 of source-geophone and its calculated velocity is just an is limited near to the surface. However, the one-sided 148
93 average of the particular velocities along its trajectory acoustic tomography can be effectively applied on a 149
94 path, without containing information about points of body if a reflecting or refracting interface with higher 150
95 acceleration or retardation. So we have a group of wave velocity exists in a lower part of the body, or in 15
96 velocities v1, v2, , vn at the relative distance intervals case that there is a positive gradient of velocity change 152
x1, x2, , xn, yielding a total time tn (an equation with with depth. In the case of a continuous positive gradient
many unknown parameters). To determine exactly these
points we need to combine a large number of sources
and receivers (matrix of equations). The set of measure-
ments is processed by specialized inversion software
solving the particular problem.
ed of velocity change, the rays of the direct waves are arc-
shaped bended due to their successive refraction, result-
ing in an increase of investigation depth. The algorithm
Rayinvr can manage such models, with gradual increase
of the velocity, and takes into account the bending of
102 158
103 Instead of the velocity, tomography (attenuation the ray-paths; therefore, it can be used in such cases. 159
104 tomography) can also describe the attenuation factor. In
105 this kind of tomography, the measurements are related 3. Data acquisition 160

106 to the amplitudes of the acoustic waves.

107 The sources and the receivers can be located either The instrumentation used was based on the 24-channel 16
108 in faced positions (crosshole tomography) or at the seismograph of the Institute of Geodynamics, EG&G 162
109 same surface (one-sided acoustic tomography). The term Geometrics StrataView R, which was able to record in 163

110 one-sided has preferentially been used in Engineering the domain of acoustic frequencies. The seismograph 164
111 literature and terminology w5x than in geophysical uses 24 bits AyD converters with a 32 kHz oversampling 165
112 research, where the method of traveltime tomography and dynamic range of 135 dB. It is capable of sampling 166
113 has been originated. In Geophysics the one-sided trav- at 32 ms (31.25 kHz), which records frequencies in the 167

114 eltime tomography is usually specified as refraction or range of 16 kHz with no aliasing. 168
115 reflection seismic tomography depending upon the The acoustic source used was a seismic hammer, and 169
116 nature of the wave arrivals used. Considering that the the receivers used had a flat response throughout the 170
117 source and the receivers are set on the ground surface, recorded frequency spectrum. The 24 receivers were 17

118 the waves generated by the source can travel to the placed at 2-m intervals along the tunnel, covering a total 172
119 receivers either directly or after critical refraction or distance of 46 m (Fig. 1a,b). Nine sources were used. 173
120 reflection on subsurface interfaces. The seismic waves The locations of the sources and the receivers are shown 174
121 have scarcely straight ray-paths since mostly the earth in Fig. 1b. The measurements were taken twice with 175
122 layers present a gradient of velocity, which increases and without a coupler at the acoustic source. When the 176
123 depth. The term seismic is mainly attributed to the coupler was used, stacking of the signal was used by 177
124 passing of waves through the earth. using three stacks. The two stacking ways were tested 178
125 The traveltime tomography usually uses a regular grid in order to have a clear and certainly identified signal. 179
126 of cells but it is also possible to use an irregular grid In order to minimize the delay time of the start of the 180
127 w6,7x. The algorithm used in the present study, Rayinvr recording, we did not implement the usual triggering 18
128 w7x, is a simultaneous refraction and reflection tomog- system of the seismograph, that was a piezoelectric 182
129 raphy program for 2-D velocity and interface structure hammer switch, but instead we preferred to use a short- 183
130 using an irregular grid. In particular, it is a ray-tracing circuit between the hammer and plate. The one cable of 184
131 algorithm accompanied by a least-square inversion rou- the triggering system was connected on the metallic 185
132 tine that uses a layered model consisted of trapezoids head of the hammer and the other one on the metallic 186
V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx 3


4 Fig. 1. (a) The location of the acoustic tomography within the tunnel of the dam. (b) Schematic layout of sources (stars) and receivers (circles)
5 within the tunnel.

187 203
coupler, giving so an instantaneous start of the recording the frequency content of each phase of wave arrival,
188 by the action of the hammer on the coupler. In recording showed that the dominant frequencies were in the order 204
189 without the coupler, we could not use the above proce- of 1.5 kHz. 205
190 dure so we led to the following method: we short- Fig. 3a shows a record processed in such a way to 206

191 circuited the triggering cables and by the hammer action enhance the P-wave (compressional wave) arrivals, 207
192 on the concrete, we cut the circuit to initiate the which come first. 208
193 recording. The section of Fig. 4 was the result of the data 209
processing by the refraction tomography inversion algo- 210
194 4. Data processing and results rithm. The tomography showed a very good fitting 211
between the observed times and the model calculated 212
195 The processing was done at a UNIX, Sun Ultra 10 times (x2s1), with an RMS error equal to 0.000072 s. 213
196 workstation, by using the programs SU 34 w15x and the This error is exceptionally small, indicating the high 214
197 inverse modeling algorithm Rayinvr w7x. SU was chosen reliability of the model to depths of 2 m. Very high 215
198 against certain commercial packages, due to its enhanced velocity values ()4500 mys) are observed along the 216
199 capabilities regarding the available amplification modes, length of the section, indicating an excellent quality 217
200 a very crucial issue when dealing with high frequency concrete (Table 1) w16x. 218
201 data. The spectral analysis that was implemented via the The small variations in velocity values are due to the 219
202 Gabor transform (Fig. 2), in order to examine separately varying degree of water saturation of the concrete. The 220
4 V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx

11 Fig. 2. A time-frequency representation of instantaneous amplitude of acoustic data via the Gabor transform. The dominant frequency of the


acoustic waves in the order of 1.5 kHz.

compressional wave velocity is increased slightly at the

presence of interstitial water. The relation between the
velocity and the porosity of the filling material is
ed 1 f 1yf
s q
V Vf Vm
225 governed by the formula of Wyllie et al. w17x: Where f is the porosity, V the acoustic velocity of the 227


18 Fig. 3. Examples of seismic records taken within the tunnel. Record (a) is specially processed to highlight the first arrivals corresponding to the
19 compressional waves, where in recording (b) the surface waves are enhanced. The first trace is at 2 m and the last one at 48 m. The vertical axis
20 is the recording time.
V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx 5

26 Fig. 4. P-wave tomography within the tunnel. All values are higher than 4.55 kmys and higher than 4.8 kmys at its largest part. The line bearing

27 is SN.
228 240
porous medium, V f the velocity of the filling fluid, and A 100% filling with water cannot be achieved in
229 Vm the relevant velocity of the non-porous medium. practice. Experiments by Popovics et al. w5x showed the 241
230 When the filling material of porous is water, V f is variation of P-wave velocity as a function of moisture 242

231 equal to 1500 mys, whereas for air V f is 330 mys. Good content of the concrete. They used a saturated concrete 243
232 quality concrete may have Vm equal to 6 kmys and sample of 39.7 kg, and measured the P-wave acoustic 244
233 porosity in the order of 2%. If we consider the extreme velocity and the Rayleigh wave velocity during the 245
234 case of all porous being filled up by water, then the drying process in an oven and the environment. Fig. 5 246
235 velocity of the material would equal to 5.66 kmys. In shows the variation of the acoustic wave velocity in 247
the case of completely dry concrete, its velocity would relation to the reducing weight of the sample due to

fall to 4.5 kmys. If we consider a 50% filling of porous
with water, its velocity would be in the order of 5 kmy
ed desiccation. From our calculations using the data of
Popovics et al. w5x, we found that the concrete they used
had porosity in the order of 2%, Vm was equal to 5.5
Table 1
33 Classification of concrete according to compressional wave velocity
Velocity P (mys) )4500 35004500 30003500 20003000 -2000

40 Concrete quality Excellent Good Doubtful Poor Very poor


47 Fig. 5. Experiment conducted by Popovic et al. w5x showed that the P-wave and Rayleigh wave velocity is increased by the moisture.
6 V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx

53 Fig. 6. Electrical tomography section at the interior of the tunnel. Values are in Ohmm.

252 272
kmys and as a result, the maximum porosity level of Fig. 6 shows the tomography interpreted model that
their sample did not reach higher than 50%. resulted from a dipoledipole array at a 2-m electrode

253 273
254 Variations of the velocity in the order of up to 400 interval. The resistivity values are in the order of 50 274
255 mys can be fully justified by the moisture content 200 Ohmm, which correspond to very wet, submerged, 275
256 variations. Thus, in the acoustic profile of Fig. 4 these splash zone concrete, according to the final report of 276
257 small variations, observed in the velocity field, could be program COST 509 for the protection of metals at their 277
258 possibly attributed to the moisture. contact to concrete w20x. 278
The determination of moisture was also attempted by
the use of electrical tomography within the tunnel.
Since the electrical resistivity of concrete depends on
porosity, moisture content, chloride penetration and
transportation mechanism w4x, electrical methods are
ed It is observed at the section that the resistivity values
down to a depth of 2 m are at relatively higher levels
compared to those at larger depths. The high resistivity
values can be attributed to low moisture content. The
increase of the values near surface is explained by the
264 used to assess the moisture and erosion hazard, in order evaporation of moisture. 284
265 to take protective measures for the concrete. The area characterized by high resistivity values, at 285
266 The resistivity values of concrete may vary between the left part of the electrical section, coincides with the 286
10 and 105 Ohmm, according to the moisture content low P-wave velocity area, found at the acoustic tomog-

267 287
268 and its nature w18,19x. The dominant mechanism of raphy section. This observation supports the following 288
269 electrical current passage through the concrete is via the hypothesis: The variations found in the velocity values 289
270 transportation of ions in the filling fluid of the porous are not due to some structural change of the material, 290
part of concrete. but due to variations of the retained moisture. At larger

271 29


59 Fig. 7. Variation of elasticity module and velocity of concrete (after Bond et al. w2x).
V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx 7

65 Fig. 8. Elasticity module tomography within the tunnel.

292 336
depths resistivity values are in the order of 40 Ohmm, The Rayleigh waves were recorded without using
293 clearly indicating the presence of moisture and saturation coupler and with no stacking. A typical example of such 337
294 of the concrete by water. a record is shown at Fig. 3b. The surface waves arrivals 338
There are additional areas along the line with very look similar to the first arrivals, since the P-wave arrivals

295 339
296 low resistivity (at depth of 3.54 m, at stations 13, 26 have been degraded after applying suitable gains. The 340
297 and 34 m), at approximately 20 Ohmm. These values P-waves can be vaguely recognized before the Rayleigh 341
298 probably imply the presence of slow water seepage, if waves arrival, not influencing the picking of them. 342
299 the occurrence of a metallic element is excluded. The main wavelength (l) of the Rayleigh waves was 343
300 The velocity variations can be attributed either to measured less than 2 m and as a result, the effective 344
moisture or to the presence of fractures, since both result
to a drop of P-wave velocity. Hence, it may be wise to
check all these points where the velocity drop is
Despite the fact that the observed P-wave velocity
ed penetration depth was at approximately 0.50.7 m.
Nevertheless, the effect of the waves to the material
could be as deep as 3 m.
The Rayleigh wave velocity can be calculated by
these measurements along the section. The measured
306 values of the concrete are slightly overestimated due to values are mainly representative to a depth of 0.50.7 350

307 its porous saturation, we may certainly argue that its m (Fig. 10). 351

308 high values are characteristic of the quality of the After taking into account the results regarding the 352
surface and compressional waves, we produced a dia-

309 material, since according to the experiment referred 353

310 previously, the velocity increase that is induced by gram of the lateral variation of the respective velocities 354

311 saturation is not more than 8% at its extreme values. ratio and the Poisson ratio s (dependent on the velocity 355

312 Therefore, we may conclude by saying that the con- ratio) (Fig. 11). The determined values are typical for 356
concrete. In highly moist areas, though the calculated

313 crete of the Marathon Dam remains of excellent quality
314 to date. Poisson ratio is less reliable, in view of a probable 358

315 From the P-wave velocity measurements, an elasticity divergence of the VRayleigh yVp ratio from its real value. 359
316 module E tomogram can be produced. A diagram of the

317 empirical relation between the elasticity module E and

318 the P-wave velocity of the concrete is presented in Fig.
319 7 w2 x .
320 The elasticity module tomography section resulting
321 from this empiric relation, is presented in Fig. 8. The

322 elasticity module values may be overestimated at the

323 moist areas, since they have been calculated on the basis
324 of the P-wave velocities. The elasticity module has
325 values of 3740 GPa, at relatively dry areas. Even these
326 smaller values are considered as very good for concrete.
327 Another important characteristic observed in Fig. 5 is
328 that the variation of VRayleigh to that of Vp, is at
329 approximately 1:2. According to the literature, the Ray-
330 leigh velocity value is at 4756% to that of the P-
331 waves, so the presence of moisture does not have a high
332 influence on the VP yVR ratio. This ratio may be corre-
333 lated to the Poisson ratio (Fig. 9). Hence, by the
334 determination of VR, we may have a general evaluation Fig. 9. The Poisson ratio has been determined from all velocity ratios 70
335 of the Poisson ratio along the tunnel. of the acoustic waves w21x. 71
8 V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx

77 Fig. 10. Lateral variation of the Rayleigh waves velocity.
360 387
At dry areas, low values of the Poisson ratio were The experiment also resulted in some general conclu-
361 observed, in the order of 0.200.25. sions about the study of concrete in highly saturated 388
362 It may have been interesting to acquire and process environment, when the one-sided acoustic tomography 389
data with shear waves within the tunnel. This was
proved unfeasible due to the large size of the source,
which involved a long beam positioned across the line
direction. We may comment at this point that the Poisson
ratio thus obtained would differ significantly from the
ed is implemented. In such an environment, the effect of
the presence of water on the increase in compressional
wave velocity measurements must be taken into account.
The lack of homogeneity in water filling of the porosity
in a body can result in small variations in measured and
368 determined values, since the shear waves are not influ- calculated quantities, such as the velocity of the com- 395
369 enced by the concrete saturation to the same degree as pressional waves, the velocity of the surface waves and 396
370 the compressional waves. the modulus of elasticity. When acoustic measurements 397

are conducted in a saturated environment, the combined 398

371 5. Conclusions use of electrical measurements is strongly recommended. 399
Electrical measurements can determine whether the pos- 400
372 The application of the one-sided acoustic traveltime sible variations in the acoustic velocity can be attributed 40

373 tomography, in combination with geoelectrical measure- to differences in water saturation or if these are related 402
374 ments within the central tunnel of the dam was highly to structural anomalies. Finally, it is possible for the 403
375 informative regarding the present condition of the dam. one-sided acoustic tomography to estimate Poisson ratio 404
376 The body of the dam is composed of high quality by taking into account the Vp yVRayleigh (compressional 405

377 concrete. This was witnessed by the high values of wave velocity to surface wave velocity ratio) instead of 406
378 acoustic waves velocity. The high quality is reflected on Vp yVs (compressional wave velocity to shear wave 407
379 the high values of elasticity module. Normal values of velocity ratio). It is worth noting that the estimation 408
380 the Poisson ratio were recorded. A large amount of with the former method is less affected by the presence 409
381 moisture is retained in the concrete pores, at the interior of water in the investigated body than the latter one. 410

382 of the dam. This has the effect of small variations on

383 the velocities of acoustic waves and in particular, it Acknowledgments 41
384 significantly lowers the values of electrical resistivity.
385 Three probable sources of slow water infiltration were The authors would like to thank EYDAP for funding 412
386 detected. the geophysical project that included our survey, and 413

83 Fig. 11. Lateral variation of the Poisson ratio.

V.K. Karastathis et al. / Construction and Building Materials xx (2002) xxxxxx 9
414 450
especially the Managing Director Mr Dionisios Xenos, w10x Sato T, Kennett BLN. Two-dimensional inversion of refraction
415 the General Director of the Networks Operation Mr traveltimes by progressive model development. Geophys J Int 451
2000;140:543 58. 452
416 Christos Aggelopoulos, the General Director of the w11x Karastathis VK, Papamarinopoulos S, Jones RE. 2-D Velocity 453
417 Planning Mr Konstantinos Kyriazis, the General Man- structure of the buried ancient canal of Xerxes: an application 454
418 ager of Water Intake and Transfer Mr Ch. Karopoulos, of seismic methods in archaeology. J Geophys 2001;47(1):29 455
419 as well as the Marathon Dam engineer Mr G. Valavanis, 43. 456
420 who assisted the project in many ways. w12x Colla C, Das PC, McCann DM, Forde MC. Sonic, electromag- 457
netic and impulse radar investigation of stone masonry bridges. 458
421 References NDTE Int 1997;30(4):249 54. 459

w13x Valle S, Zanzi L, Binda L, Saisi A, Lenzi G. Tomography for 460
NDT applied to masonry structures: sonic andyor EM methods.

422 w1x McCann DM, Forde MC. Review of NDT methods in the 461
423 assessment of concrete and masonry structures. NDT&E Int Proceedings of the Second International Arc Bridge Confer- 462
424 2001;34(2):71 84. ence. Venice, 1998:243252. 463
425 w2x Bond LJ, Kepler WF, Frangopol DM. Improved assessment of w14x Schuller MP, Berra M, Atkinson R, Binda L. Acoustic tomog- 464
426 mass concrete dams using acoustic travel time tomography. raphy for evaluation of unreinforced masonry. Sixth Confer- 465
427 Part Itheory. Constr Build Mater 2000;14(3):133 46. ence on Structural Faults and Repair 1995;3:195 200. 466

428 w3x Kepler WF, Bond LJ, Frangopol DM. Improved assessment of w15x Cohen JK, Stockwell JW. CWPySU, seismic unix release 30. 467
429 mass concrete dams using acoustic travel time tomography. A free package for seismic research and processing. Colorado 468
430 Part IIapplication. Constr Build Mater 2000;14(3):147 56. School of Mines, CO, USA: Center for Wave Phenomena, 469
431 w4x Polder RB. Test methods for on site measurement of resistivity 1997. 470
432 of concretea RILEM TC-154 technical recommendation. w16x Whitehurst E. Soniscope tests concrete structures. J Am Concr 471
433 Constr Build Mater 2001;15(23):125 31. Inst 1951;47:433 44. 472
434 w5x Popovics JS, Song W, Achenbach JD, Lee JH, Andre RF. One- w17x Wyllie MRJ, Gregory AR, Gardner LW. Elastic wave velocities
sided stress wave velocity measurement in concrete. ASCE J
Eng Mech 1998;124:12.
w6x Vesnaver A, Bohm G, Galuppo P. Staggered versus adapted
grids for the joint 3-D inversion of surface and well data.
Proceedings of the 69th Annual Conference on International
Management: Society of Exploration and Geophysics.
ed in heterogeneous and porous media. Geophysics 1956;21:41
w18x Gjrv OE, Vennesland , El-Busaidy, AHS. Electrical resistiv-
ity of concrete in the oceans. Ninth Annual Offshore Technol-
ogy Conference, Paper 2803. Houston, 1977.
w19x Tuutti K. Corrosion of steel in concrete. Stockholm: CBI, 1982. 479
441 1999:17931796.
442 w7x Zelt CA, Smith RB. Seismic traveltime inversion for 2-D (p. 468). 480
443 crustal velocity structure. Geophys J Int 1992;108:16 34. w20x COST 509. Corrosion and protection of metals in contact with 481
444 w8x OLeary DM, Ellis RM, Stephenson RA, Lane LS, Zelt CA. concrete. In: Cox RN, Cigna R, Vennesland O, Valente T, 482

445 Crustal structure of the northern Yukon and Mackenzie Delta, editors. Final report, EUR 17608 EN, ISBN 92-828-0252-3. 483
446 northwestern Canada. J Geophys Res 1995;100:9905 20. Brussels: European Commission, Directorate General Science, 484
447 w9x Zelt CA, White DJ. Crustal structure and tectonics of the Research and Development, 1997. (p. 148). 485
448 southeastern Canadian Cordillera. J Geophys Res w21x Sheriff RE, Geldart LP. Exploration seismology, vol. 1. Cam- 486

449 1995;100:24255 73. bridge University Press, 1982. 487