SHEAR-WAVE VELOCITY PROFILE FOR HOLOCENE SEDIMENTS MEASURED FROM MICROTREMOR ARRAY STUDIES, SCPT, AND SEISMIC REFRACTION Michael W. Asten 1
W.R. (Bill) Stephenson 2 and Peter N. Davenport 2
1
Centre for Environmental and Geotechnical Applications of Surface Waves (CEGAS), School of Geosciences, Monash University, Melbourne, Vic. Australia
[email protected] 2 Institute of Geological and Nuclear Sciences (IGNS), Lower Hutt, New Zealand.
Paper accepted for publication in the Journal of Engineering and Environmental Geophysics, expected date Sept 2005.
Abstract The microtremor method was trialed on soft Holocene sediments overlying glacial gravels near the Hutt River, Wellington New Zealand. A triangular four-station array of radius 40 m, processed using a modified spatial autocorrelation (SPAC) method was sufficient to establish shear velocity for sediments and gravels, and a thickness of 21.8 m for the soft sediments. The interpretation was performed blind to geological and geophysical data, and subsequent comparison with seismic cone penetrometer and seismic refraction data shows the accuracy from the SPAC interpretation to be better than 5%. Interpretation of horizontal:vertical spectral ratio data assists with estimates of thickness of the glacial gravels (order hundreds of meters) but accuracy in thickness is poor, perhaps due to uncertainty in the 2D nature of the base of the gravel-filled glacial valley.
Introduction The Greater Wellington region of New Zealand is an area of high seismic hazard due to the presence of active fault lines passing through its urban areas, and due to the thickness of glacial and alluvial sediments which underlay a large proportion of the urban areas. The importance of characterizing the seismic response of these sediments has led to programs over recent decades to map the sediments with surface seismic methods, and measure shear-velocity profiles with seismic cone penetrometer methods (Robertson et al 1986). Increasing restrictions on invasive studies due to the risk of disturbance to aquifers in the sediments have increased the importance of developing alternative non-invasive methods for measurement of the shear-wave velocity profile. This paper summarizes results of a trial of the microtremor array method (MAM) at McEwan Park located close to where the Hutt River discharges into Wellington harbour. (see Figure 1). The site was selected by one of the authors (WS) as a suitable trial site which had prior seismic refraction data (using both P and converted S-waves) and SCPT data. The microtremor data was interpreted by another author (MWA) “blind”, ie. no geological or geophysical data was made available until the completed interpretation of microtremor data as presented here was placed on file at the IGNS New Zealand.
The field area McEwan Park is a grassed recreation area situated close to where the Hutt river discharges into the sheltered waters of Wellington harbour. For the last 500,000 years, periodic vertical movements on the 3km distant Wellington fault, and on the 20km distant West Wairarapa fault have caused the land at McEwan Park to rise 2 meters every 2500 years (West Wairarapa fault) and fall one meter every 600 years (Wellington fault). These movements have resulted in interbedded sands, silts and clays being deposited at the site during Holocene times, as the floor of the Hutt valley has been raised and lowered. Pre 1855 this area was an estuary of the Hutt River, but after the latest uplift a sand bar developed. Below the Holocene materials lie stiffer materials corresponding to four glacial and four interglacial regimes, deposited from 1.8Ma ago, and reaching to a maximum depth of 300m. Immediately beneath the Holocene deposits is a 25m layer of gravels from the Otira glaciation, and these constitute the Waiwhetu aquifer. The other glacial deposits are sandy gravels, and constitute less important aquifers. Extensive urban development has led to many site investigations in the Hutt valley and consequently many borehole logs for the area are available. Furthermore the need to understand the aquifers that supply the nearby urban area has led to three boreholes being sunk to bedrock. Thus there is a wealth of material describing the geological setting of the Hutt valley. Unfortunately little of this material can be interpreted in terms of shear wave velocities. The sources can be accessed by following the references in Begg and Mazengarb (1996) and in Dellow et al (1992). A Seismic Cone Penetration Test at McEwan Park (Stephenson and Barker, 1992) showed that soft soils deduced to be of Holocene age occupy the top 21m, and a borehole encountered bedrock at 175m depth at a site 800m east of McEwan Park. However at a site 400 m northwest of McEwan Park, bedrock was not encountered at 200 m depth. Phase velocities from microtremors
The MAM used here for estimating shear-wave velocity was proposed by Aki (1957) and has been reviewed by many authors (eg. Tokimatsu, 1997; Asten, 2001; and Okada, 2003). Microtremors are the background movement of the earth attributable to non-seismic sources. In the frequency band of interest in this study (1 to 30 Hz), sources are principally cultural noise such as vehicle traffic and industrial machinery. For studies in metropolitan areas, the MAM is especially useful as the seismic noise which degrades active seismic methods (such as seismic reflection and refraction surveys) provides a plentiful source of energy for passive seismic methods. The microtremor seismic energy detected by vertical-component geophones propagates principally as fundamental-mode Rayleigh (surface) waves. The phase velocity of propagation is dependent mainly on shear-wave velocity (and less so on compressional velocity and density) of the underlying regolith. The velocity of propagation is a function of period; long periods have long wavelengths which are affected by rock properties deeper in the earth, while short periods have short wavelengths and are affected by rock properties at correspondingly shallower depths. In the MAM, we measure phase velocity over a range of frequencies and invert these to a layeredearth model by iterative fitting of observed and modeled dispersion data. Numerous seismic sources contribute energy to the MAM at any instant of time, hence the direction of wave propagation is undefined. Consequently, standard ray-path methods used in conventional seismology are not appropriate for this technique. Instead, we use spatiallyaveraged coherencies (SPAC) computed using a small circular seismic array as shown in Figure 2. The spatially averaged coherency for a single-mode surface wave observed with a circular seismic array is given by ave c(f) = J0 (kr) = J0 (2 π f r / V(f) ),
(1)
where ave c(f) is azimuthally-averaged coherency, f = frequency, J0 is the Bessel function of zero order, k is the scalar wavenumber, V(f) is the required phase velocity dispersion curve, and r is the station separation in the circular array (Aki, 1957). The seven-station hexagonal array shown in Figure 2 allows estimates of SPAC over three different spatial separations r1 = r (the six radii of the hexagon), r2 = r (the six circumferential station separations on the perimeter), r3 = 1.7r (the off-diagonal separation) and r4 = 2r (the diagonal of the circle). Estimating coherencies simultaneously at three separations theoretically gives sufficient redundancy to identify frequencies where energy is propagating in two modes simultaneously (such as the fundamental and first higher modes), and to solve for the two phase velocities and energy partition coefficient (Asten, 2001), hence we use the term multi-mode SPAC (MMSPAC) for this approach. The classical method of analysis is to fit azimuthally-averaged coherencies at each frequency to a Bessel function curve, and thus obtain a value for phase velocity V(f) (Okada, 2003; Ohori, 2002). The phase velocity data is then fitted to modeled phase-velocity dispersion curves either by iterative forward modeling, or by inversion algorithms. In this work we use an alternative procedure following Asten et al. (2002) and Asten et al (2004) whereby we compute modeled phase velocity dispersion curves, compute modeled coherency vs frequency curves via eqn (1), and then fit observed and model coherencies by iterative forward modeling. The modeled dispersion curves are obtained for layered earths by using routines published by Herrmann (2001). This approach is similar to that used by Chouet et al (1998), proves more robust than the classical SPAC methodology, and greatly facilitates recognition of multi-mode Rayleigh-wave propagation if present.
Coherency spectra shown in Figures 3 and 4 shows that microtremor data acquired at this site show no evidence in the frequency band 2.5 to 15 Hz for significant departures from fundamental-mode Rayleigh-wave propagation. Instrumentation and data processing The site is a turf sports ground. A set of seven Mark L4C3D three-component geophones was used for microtremor detection, with the geophone feet pressed firmly into the turf. Signals from each geophone were recorded at 200 samples/sec and 12-bit resolution by a three-channel EARSS (Equipment for the Automatic Recording of Seismic Signals ) recorder, described by Gledhill et al (1991), with each recorder synchronized separately to GPS time. Approximately 30minutes of data was acquired with each array used at this site. For processing the data, synchronized records of length 5 minutes, free of obvious spikes from near-by road traffic, were weighted with a Hanning bell and transformed to the frequency domain. A coherency matrix for the 21 possible pairs of vertical-component stations was constructed, using smoothing over 40 frequency values of the complex cross-spectrum. Selected terms of the coherency matrix were averaged to give the azimuthally-averaged coherency (SPAC) spectra corresponding to station separations r1, r2, r3 and r4 outlined above. Results In the example discussed here, field data was acquired with two separate circular arrays of radii 22 m and 40 m. Figure 3 shows SPAC spectra for a data sample acquired with the 22 m radius array, together with the modeled SPAC spectra for the fundamental and 1st-higher Rayleigh modes for the best-fit layered earth, shown in Table 1 and Figure 5. The band of useful frequencies extends from 2.5 Hz to 18 Hz. Figure 4 shows SPAC spectra for a data sample acquired with the 40 m radius array; in this case one of the outer recording units failed and so for processing purposes the array was reduced to a four-station triangular array which allows SPAC spectra to be computed at only two station separations, being the radius 40 m, and the triangle side length 69 m. This larger array shows a closely similar useful frequency range and in hindsight we may conclude that at this site the use of the larger triangle array alone is sufficient to obtain the desired Vs profile. For purposes of comparison with alternative shear-wave techniques, Figure 5 shows the phase-velocity dispersion curves for the best-fit model, together with observed phase velocities computed from the field SPAC spectra of Figure 4. The interpreted model shows a total thickness of 21.8 m of soft sediments overlying hundreds of meters of firm sediments; these are interpreted to be Holocene sands/silts and Pleistocene glacial tills respectively. Figures 7, 8 and 9 illustrate a basic sensitivity analysis on the model. Figure 7 shows that shear-velocity variations of +-10% in the Holocene sediments are not admissible by the data and hence the sensitivity to Vs in the soft sediments is of order 5%. Figure 8 shows that the data (in particular the SPAC curve zero crossing at 2.3 Hz) places a lower bound of order 550 m/sec on Vs for layer 5 (glacial gravels) but does not clearly resolve an upper bound. Figure 9 shows the effect on modeled SPAC of reducing the thickness of the gravels (h5) to 200 m, and to 1 m; the field data does not differentiate between h5= 400 m or 200 m, but weakly indicates that 1 m is too thin (a poorer model). We conclude from Figures 8 and 9 that this SPAC data in isolation gives only weak evidence for the existence of the underlying glacial gravels, and does not resolve their thickness. We have attempted to use the horizontal:vertical spectral ratio (HVSR) method as proposed by Nakamura (1989) and used by Tokimatsu (1997) and Satoh et al (2001) to provide an estimate of total sediment thickness. Figure 10 shows that the site has a broad maximum in the vicinity of 0.3
Hz; this was interpreted while “blind” to geology as a low-frequency site resonance and requires a layer of Vs=550 m/sec with thickness 400 m in order to produce a modeled particle motion ellipticity (Figure 10) having a maximum near the observed 0.3 Hz HVSR maximum. However the HVSR peak is much broader than is usual for a layered earth, and the thickness estimate of 400 m from HVSR is much greater than the likely thickness of order 200 m based on geological data from two boreholes located within 800 m east and west. We are unable to say whether these discrepancies are due to poor instrument response below 1 Hz, to 2D or 3D geological structure, or other factors unknown. Layer velocities from scpt Shortly after the announcement of the SPCT method (Robertson et al, 1986) IGNS implemented a preliminary version of the method, which used fast paper recording instead of digital recording to log arrivals. As a consequence of this early arrangement the only data retained for the McEwan Park site was a table of first arrivals versus depth. These data are plotted in Figure 11, and it shows a 21 m layer of material having a shear wave velocity of 160 m/s. The cone hit gravel at depth 21.0 m (logging at 0.2 m intervals). Full results for other parts of the Hutt valley are given by Stephenson and Barker (1992) who also show that the 21 m layer at McEwan Park is broken into two parts with different penetration properties, reflecting a 10.6 m layer of alluvium over an 10.4 m layer of estuarine mud. Layer velocities from seismic refraction Acknowledging the importance of the McEwan Park site, IGNS carried out explosion source seismic refraction studies as reported by Stephenson and O’Reilly (2002). Two sets of arrivals in particular were noted. Figure 12 shows radial arrivals corresponding to converted waves which have traveled at 525m/s as Sv waves in the glacial gravels, and Figure 13 shows vertical arrivals corresponding to direct P waves which have traveled at 1870m/s in the top 21m layer. The shear-wave velocity value of 525m/s given in Figure 12 from seismic refraction data is in striking agreement with the value of 550m/s given in Figure 6, obtained between 21.8m and 422m using SPAC interpretation without prior knowledge of the refraction data. Conclusions In the case of homogeneous soft Holocene sediments overlying glacial gravels at McEwan Park, MMSPAC interpreted blind to independent geological and seismic data provides accurate values of shear wave velocity in both the soft layer and in the underlying gravels, as well as an accurate value of the thickness of the soft layer. Accuracy for the Holocene sediments to 21 m depth is better than the figure of 5% previously claimed for this method by Asten et al (2004). Acknowledgments Funding for the data acquisition was provided by Flagstaff GeoConsultants Pty Ltd, Melbourne, and the IGNS, Wellington. Interpretation methodologies for this data were developed by Michael Asten who is supported by the U.S. Geological Survey (USGS), Department of the Interior, under USGS award number 04HQGR0030. We thank Hutt City Council for allowing access to McEwan Park, and for providing the base map from which Figure 1 is derived. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either express or implied, of the U.S. Government. References
Aki, K., 1957, Space and time spectra of stationary stochastic waves, with special reference to microtremors: Bull, Earthq. Res. Inst, 35, 415-456. Asten, M.W., 2001, The Spatial Auto-Correlation Method for Phase Velocity of Microseisms – Another Method for Characterisation of Sedimentary Overburden: in Earthquake Codes in the Real World, Australian Earthquake Engineering Soc., Proceedings of the 2001 Conference, Canberra, Paper 28. Asten, M.W., Lam, N., Gibson, G. and Wilson, J., 2002, Microtremor survey design optimised for application to site amplification and resonance modelling, in Total Risk Management in the Privatised Era, editted by M Griffith, D. Love, P McBean, A McDougall, B. Butler, Proceedings of Conference, Australian Earthquake Engineering Soc. , Adelaide, Paper 7. Asten, M.W., 2004, Passive seismic methods using the microtremor wave field for engineering and earthquake site zonation: 74th Annual Meeting of the SEG, Denver, Oct 2004, Session NSG-1. Asten, M.W., Dhu, T., and Lam,N., 2004, Optimised array design for microtremor array studies applied to site classification; observations, results and future use: Paper 2903, Conference Proceedings of the 13th World Conference of Earthquake Engineering, Vancouver, Aug 1-6. Asten, M.W., and Boore, D.M., 2005, Comparison of shear-velocity profiles of unconsolidated sediments near the Coyote borehole (CCOC) measured with fourteen invasive and noninvasive methods, in Asten, M.W. and and Boore, D.M. (editors), Blind comparisons of shear-wave velocities at closely-spaced sites in San Jose, California, Proceedings of a Workshop held at the US Geological Survey, Menlo Park, May 3, 2004. US GS Open File Report 20005/xxxx, in review. Begg, J.G.; Mazengarb, C.,1996, Geology of the Wellington area : sheets R27, R28, and part Q27, scale 1:50,000.Lower Hutt: Institute of Geological & Nuclear Sciences. Institute of Geological & Nuclear Sciences geological map 22. 128 p. + 1 fold. Map. Chouet, B., De Luca, G., Milana, G., Dawson, P., Martini, M., and Scarpa, R., 1998, Shallow velocity structure of Stromboli Volcano, Italy, derived from small aperture array measurements of Strombolian tremor. Bull. Seism.Soc. Am. 88, 653-666. Dellow, G.D.; Read, S.A.L.; Begg, J.G.; Van Dissen, R.J.; Perrin, N.D., 1992, Distribution of geological materials in Lower Hutt and Porirua, New Zealand : a component of a ground shaking hazard assessment.Bulletin of the New Zealand National Society for Earthquake Engineering, 25, 332-344. Gledhill, K.R.; Randall, M.J.; Chadwick, M.P., 1991, The EARSS digital seismograph : system description and field trials. Bulletin of the Seismological Society of America, 81, 13801390. Herrmann, R.B., 2001, Computer programs in seismology - an overview of synthetic seismogram computation Version 3.1, Department of Earth and Planetary Sciences, St Louis Univ. Nakamura, Y., 1989, A method for dynamic characteristics estimation of subsurface using microtremors on the ground surface: Quarterly reports of the Railway Technical Research Institute Tokyo, 30, 25-33. Ohori, M., Nobata, A., and Wakamatsu, K., 2002, A comparison of ESAC and FK methods of estimating phase velocity using arbitrarily shaped microtremor arrays. Bull. Seism. Soc. Am. 92, 2323-2332. Okada, H., 2003, The Microseismic Survey Method: Society of Exploration Geophysicists of Japan. Translated by Koya Suto, Geophysical Monograph Series No. 12, Society of Exploration Geophysicists, Tulsa. Roberts, J., and Asten, M.W., 2004, Resolving a velocity inversion at the geotechnical scale using the microtremor (passive seismic) survey method: Exploration Geophysics, 35, 14-18. Roberts, J., and Asten, M.W., 2005, Estimating the shear velocity profile of Quaternary silts using microtremor array (SPAC) measurements: Exploration Geophysics, 36, 34-40. Robertson, P. K., Campanella, R. G., Gillespie, D., Rice, A., 1986, Seismic CPT to measure in situ shear wave velocity: Journal of Geotechnical Engineering, 112, 791-803.
Satoh, T., Kawase, H.,Iwata, T., Higashi, S., Sato, T., Irikura, K., and Huang, H., 2001, S-wave velocity structure of the Taichung basin, Taiwan, estimated from array and single-station records of microtremors: Bull. Seism. Soc. Am. 91, 1267-1282. Stephenson, W.R.; Barker, P.R., 1992, Evaluation of sediment properties in the Lower Hutt and Porirua areas by means of cone and seismic cone penetration tests: Bulletin of the New Zealand National Society for Earthquake Engineering, 25, 265-285. Stephenson, W.R.; O'Reilly, C.W., 2002, Using a moving source to generate Rayleigh waves.Lower Hutt: Institute of Geological & Nuclear Sciences Limited. Institute of Geological & Nuclear Sciences Science Report 2002/15. 18 p. Tokimatsu, K., 1997, Geotechnical site characterization using surface waves, in Earthquake Geotechnical Engineering, editted by Ishihara. Balkema, Rotterdam.
TABLE 1 Layered-earth model from interpretation of McEwan Park microtremor data, For layer parameters thickness(H), compressional and shear velocity (Vp, Vs) and density (RHO). # 1 2 3 4 5 6 7 8
H Vp Vs RHO (m) (m/sec) (m/sec) (t/m3) 2 500 130 1.8 2.8 1700 140 2.0 8 1700 160 2.0 9 1700 160 2.0 400. 1700 550 2.4 1040. 3880 2230 2.4 1800. 4630 2680 2.4 inf 6040 3490 2.8
GEOLOGY Holocene silts and mud Glacial gravels Basement (unresolved)
Notes: Vp and rho values are guesses based on probable depth to water table for sands and gravels (layers 1-5). For basement rock (layers 6-8) Vp and Vs are not resolved by these measurements but are included as assumed reasonable values for hard rock.
r4 =2 r1
r1 = 50 m
r3=1.7 r1 r2 = r1 Fig. 2. A hexagonal array of 7 stations, for SPAC processing with four inter-station distances, allowing recognition of dual modes of phase velocities at each frequency.
Fig. 1. Location of McEwan Park, Lower Hutt, NZ.
(a)
(a)
(b)
(b)
(c)
Fig. 4. SPAC coherency spectra for a 5minute data segment acquired with the 40 m radius four-station triangular array. (a) r1=40 m, (b) r2=69 m. Black line is SPAC spectrum of field data. Dashed (doted) lines are modeled SPAC spectrum for the fundamental (first higher) Rayleigh modes using the shear-velocity profile from Fig. 6.
(d)
Fig. 3. SPAC coherency spectra for a 5minute data segment acquired with the 22 m radius hexagonal array. (a) r1=22 m, (b) r2=22 m, (c ) r3=38 m, (d) r4=44 m. Black line is SPAC spectrum of field data. Dashed (dotted) lines are modeled SPAC spectra for the fundamental (first higher) Rayleigh mode, using the shear-velocity profile from Fig. 6.
Fig. 5. Computed dispersion curves for fundamental and first higher Rayleigh modes, for the layered earth model shown in Figure 6. Superimposed points are phase velocity estimates from the field coherency curves shown in Figure 4.
McEwan Park Vs Velocity (m/sec) 0 0
100
200
300
400
500
600
-5
-10
depth (m)
-15
-20
-25
-30
-35
-40
Fig. 6. McEwan Park final layered-earth model derived from SPAC data. Densities and Vp are assumed (see Table 1). Values for Vs and h are obtained by iterative forward modeling.
(a)
(a)
(b)
h5=1m h5=200 m
(b)
h5=400 m Fig. 7. Field and model SPAC data as for Figure 4. The dashed line is the final model. The two overprinted models (thin black lines) show the final model with +-10% perturbation of Vs4 (Holocene muds).
(a)
Vs5=770 m/s Vs5=550 m/s
(b)
Vs5=440 m/s Fig. 8. Field and model SPAC data as for Figure 4. The dashed line is the final model. The two overprinted models (thin black lines) show the final model with -20% & +40% perturbation of Vs5 (glacial gravels).
Fig. 9. Field and model SPAC data as for Figure 4.. The dashed line is the final model. The two overprinted models (thin black lines) show the final model with reduction of thickness of glacial gravels (h5) to 200 m, and to 1 m. The data does not resolve thickness of h5 in the range 200-400 m, but the observed H/V maximum around 0.3 Hz (Fig. 10) requires h5 to be of order 400m in order to get a modeled H/V maximum at 0.3 Hz.
Fig. 10. McEwan Park: (TOP) field HVSR. X-axis is frequency in Hz. (Bottom) Modeled ellipticity for fundamental-mode (dashed line) and 1st higher mode (dotted line) Rayleigh waves for the layered earth model of Figure 6.
Distance from shot point (m)
0 10 20 30 40 50 60 70 80 90
100 110 120 130
140 150 0
Figure 11. Seismic CPT profile at McEwan Park. Dashed line corresponds to a shear wave velocity of 160m/s.
0.15
Time (sec) Figure 13. Vertical arrivals from explosion seismology. Dashed line corresponds to P waves travelling at 1870m/s in the top 22m thick layer of Holocene material.
Figure 12. Radial arrivals from explosion seismology. Dashed line corresponds to waves travelling as Sv at 525m/s in gravels below 22m.