Site characterization by seismic interferometry method - CiteSeerX

Geotechnical and Geophysical Site Characterization – Huang & Mayne (eds) © 2008 Taylor & Francis Group, London, ISBN 978-0-415-46936-4

Site characterization by seismic interferometry method T. Matsuoka & K. Onishi Department of Civil and Earth Resources Engineering, Kyoto University

K. Shiraishi Japan Petroleum Exploration Co., Ltd

T. Aizawa Suncoh Consultants, Co., Ltd

ABSTRACT: Seismic interferometry is a new signal processing technique for generating Green’s function by applying cross-correlation operations to observed seismic data at different receiver locations. This method can be applied to imaging subsurface structures with observing underground noise in natural or in social life. We carried out a field survey using Digi-pulse, air-knocker hit, and a truck running vibrations in tunnel. A survey line with 96 receivers was located on the top of hill just above the tunnel. Obtained subsurface images are clear enough to identify subsurface structures. Our experimental results strongly support applicability of the seismic interferometry to subsurface imaging from the passive data of the noise by social activities. The interferometric imaging will be useful in site characterization for applications in civil engineering, environment, and disaster preventions.

1

INSTRUCTION

Reflection seismology has been widely applied to investigate subsurface structures not only in the oil and gas industry but also for the purpose of civil engineering, environmental and disaster prevention (McCann et al., 1997, Brouwer & Helbig, 1998). These can be called the site characterization using seismic survey. However, this method requires artificial explosive sources such as dynamite on the ground surface in order to produce the elastic waves which propagate underground and return the geological information to the surface.A great deal of research has been attempted in order to develop appropriate artificial sources for the data acquisition environments. In this paper a methodology of reflection seismology without using artificial sources is proposed with a field example. Including small shakes that cannot be felt by humans, underground motions occur almost daily due to the activity of earthquakes and volcanoes. The idea of using these underground motions as sources for reflection seismology can be realized using the theory of seismic interferometry (Lobkis, and Weaver, 2001, Wapenaar, 2003, Wapenaar, and Fokkema, 2006, Shiraishi, et al., 2007). In the theory of seismic interferometry, the Green’s function of the seismic wave propagation between

two receiver positions can be reconstructed by taking the cross-correlation of observed records at these two different places. This basic theorem suggests that we can synthesize reflection seismic data between two positions on the ground surface through the cross-correlation of transmitted seismic data from an underground seismic source which is observed at these two places. In this case there are no restrictions regarding the cause of these seismic vibrations. General relations between the reflection and transmission response were studied by Claerbout (1968) for the onedimensional case and by Wapenaar et al. (2004) for the three-dimensional case. In this study, we try to image the subsurface structures from the aftershocks observed at theTottori-seibu earthquake in Japan. We observed 22 aftershocks with 255 receivers along the road. Then pseudo shot data were synthesized at each receiver location. Conventional data processing was applied to the synthesized seismic data and finally a seismic section was obtained with which we could interpret subsurface structures. In conclusion, this technique is easily applied to other seismological events and it is useful for geophysical surveys using the continuous monitoring system. This methodology will become more important for disaster prevention and Earth science in the future. Under the ground, there always exist various kinds of vibrations

891

Figure 2. Two layers model with concave structure for numerical simulation.

responses can be expressed as follows (Wapenaar, 2003).

Figure 1. Conceptual diagrams of seismic interferometry.

generated by social life, such as the running of trains, subways, or cars and civil constructions and factories in addition to natural noises such as earthquakes or volcanic activities. We adopt the “Seismic interferometry” to obtain the reflection responses. The method intends to image the subsurface geological structure using the noise as a transmission wave field observed on the surface and to simulate the pseudo shot records as many as the number of the receivers corresponding to data of reflection seismic survey.

2

SEISMIC INTERFEROMETRY

The relationship between the transmission response and the reflection response in the one dimensional multilayer structure was developed by Claerbout (1968), and this idea was extended to the three dimensional case by Wapenaar (2003). Once the seismic wave fields from buried sources are observed at two places A and B on the surface of ground, we can simulate the pseudo reflection wave field which corresponds to the recorded seismogram at B of the source generated at A (Fig. 1). The expression of the relationship of the transmission response T and the reflection response R due to a source at xi is as follows (Wapenaar 2003).

Where T (xA , xi , t) and T (xB , xi , t) are the transmission wave fields at points A and B from the subsurface source point xi respectively. R(xA , xB , t) is the reflection wave field corresponding to the receiver at xA and the source is placed at xB . The symbol ∗ denotes convolution. Using this relationship, the reflection wave field can be simulated by the cross-correlation of both transmission wave fields. The first term of left-hand side of equation (1) is causal part and the second term is the non-causal part. The second term of the righthand side of equation (1) expresses the summation of the correlation results for each source in case that two or more buried sources exist. When there exist random vibration in the subsurface, the relationship between the reflection responses and the transmission

This Eq. (2) is similar to Eq. (1), however, crosscorrelation appears only once between observed transmission responses Tobs (xA , t) andTobs (xB ,t). When we passively observe underground wavefield for the seismic interferometry, it is necessary to acquire the data as long as possible in order to improved signal to noise ratio. In equation (1) and (2), the combination of receivers, xA and xB can be exchanged. It means that we can simulate the shot records at each receiver position as many as the number of receivers deployed on a survey line, which are correspondent to the shot records in conventional reflection seismology. This is a great advantage of this method. After we obtain the simulated shot records, we can process them according to the data processing in the conventional reflection seismology. 3

NUMERICAL SIMULATION

We show here some numerical examples to simulate shot records by cross-correlation from transmission records due to moving impulsive sources and random vibration in the ground. We use a model with a simple concave (syncline) structure, whose velocity is 1,500 m/s in an upper layer and 2,500 m/s in a lower layer (Fig. 2). 71 receivers are located on the surface from x = 2,600 m to x = 5,400 m with 40 m spacing. In simulation of moving sources, Ricker wavelet (25 Hz) is used as source wavelet at each source position; 201 shots with 30 m interval between x = 1,000 m and x = 7,000 m, at z = 500 m in depth. In simulation of random vibration, band-limited random noise is generated at each source position. Then, we simulate shot records by cross-correlation from the observed transmission records. For the moving impulsive sources, we obtain the simulated shot record by summing the all cross-correlation results of each transmission record. For the random vibration observation, we use 30-minute data. Figure 3 shows the simulated shot records calculated by crosscorrelation from moving impulsive sources (a), and random vibration (b), and true reflection response (c)

892

Figure 3. Simulated shot records from moving impulsive sources (a), and from random noise sources (b), and true reflection response (c) when a source is located at the center of the survey lines.

Figure 5. PC-based recording system.

Figure 6. Moving sources. From left to right: Digi-pulse, air-knocker, 2t truck.

4.2 Data recording system

for comparison, when the source is located at the center of the line (x = 4,000 m). We can find direct waves and reflections and their multiple reflections are reconstructed correctly. If we obtain simulated shot records at each receiver position, we can image the subsurface structures by according to the conventional processing. These theoretical simulations support our field experiment with moving impulsive source in the ground.

For the effective seismic interferometry, the long time recording data with many receivers is desirable. For this field experiment, we newly developed the PCbases seismic recording system (Fig. 5) which can continuously record the sampled data of 96 receivers simultaneously for about one hour. The output from the each geophone enters the A/D converter of the PC and the data are stored on a hard disk. The PC-based data acquisition system is controlled by LabView. This PC-based recording system can record the data continuously for arbitrary time period with 1000 Hz sampling frequency. After starting the recording system, the source in the tunnel begins to move and stop at the specified period of time or the specified number of the shot times.

4

4.3 Data acquisition

Figure 4. Geology, topography, and survey line.

4.1

FIELD TEST WITH ARTIFICIAL SOURCES IN TUNNELS Survey layout

We carried out a field test with underground moving sources in a hilly area where there exist two tunnels under the ground. The geological structures of the test field and survey lines are shown in Fig. 4. In this area, it is known that the alternation of mudstone of a single inclination which is parallel to the east side slope of a hill by a prior geological survey. The driveway tunnel and the sidewalk tunnel pass in the ground. These tunnels are used as a road of the moving sources. A survey line (Line-1) on which 96 receivers with an interval of 5 m were deployed, was located on the hill just above a sidewalk tunnel. We also acquired the regular seismic data for the comparisons with sections by seismic interferometry.

As the seismic source, following three kinds of moving sources were used both in a driveway tunnel and a sidewalk tunnel. Figure 6 shows a picture of each source, a Digi-pulse, an air-knocker, and a 2-ton truck. The Digi-pulse and air-knocker hit the surface of the tunnel road at equal intervals (every 2.5 m–5 m) in the driveway tunnel and the sidewalk tunnel respectively, moving from the entrance to the exit of the tunnels. The truck kept running in the driveway tunnel intently and drove 10 times round trips. The record lengths of each source is 1,840s (Digi-pulse), 2,140s (air-knocker) and 1,600s (truck). In actual recording, we cannot apply the recording trigger because of no information of the seismic source movement. Since a recording trigger was not applied during the data acquisition, the first break of each record is not the same. However, the theory guarantees the unnecessity of first break

893

Figure 7. Simulated shot records. From left to right: Digi-pulse, air-knocker, 4t truck.

reflection seismology (with 72 channels) are shown in Fig. 8. From Fig. 8, a geological structure parallel to the surface inclining to the east side can be clearly visualized. This structure crosses the west side slope of the hill. The results from the seismic interferometry are identical to the result of the reflection seismology. This fact strongly supports the validity of the method of the passive record observation, seismic interferometry. We obtained subsurface image of the hill by using the hitting type sources and also by truck running vibrations. The long data acquisition may enable to improve the signal-to-noise ratio. This result shows the applicability of seismic interferometry to field data not by using the artificial sources such as the dynamite explosive. We can also presume to estimate underground structure from the vibrations produced in a social activity, and this suggests that this technique can play an active part in a many field. 5

CONCLUSION

matching of each record because of cross-correlation operation.

In this paper, a field experiment towards practical use of the seismic interferometry was done. Since this method requires a long record length to improve the S/N ratio, we developed a PC-based seismic recording system with 96 channels for this purpose. We obtained subsurface images of the hill by using the hitting type sources and also by using truck running vibrations. This result shows the applicability of the seismic interferometry to estimate underground structure from the vibrations produced in a social activity. The method discussed here is a newly proposed reflection seismology technique by using the micro seismic wave field positively without artificial sources, which has so far been considered to be a noise in reflection seismology.

4.4

REFERENCES

Figure 8. Stacked Sections. From left to right: (a) Digi-pulse, (b) air-knocker, (c) 4t truck, and (d) surface reflection survey.

Data processing

In the data processing, in order to remove the noises on the measurement and PC-based recording system, a median filter and a band pass filter were applied at first processing stage. Then, simulated shot records are generated by cross-correlation of the observed records, and the header information was added and finally converted to the SEG-Y formatted data. Figure 7 shows the example of the simulated shot records obtained from each seismic source. Each shot record in Fig. 7 can show us relatively clear reflection event corresponding to the geological boundary. In order to image the subsurface structures, the conventional data processing was applied to the simulated shot records. The processed stacked sections by the seismic interferometry and by the conventional

Claerbout, J.F. (1968): Synthesis of a layered medium from its acoustic transmission response, Geophysics, 33, 264–269. Lobkis, O. and Weaver, R. (2001): On the emergence of the Green’s function in the correlations of a diffuse field, Journal of the Acoustical Society of America, 110, pp. 3011–3017. Shiraishi, K., Matsuoka, T., Matsuoka, T., Tanoue, M. and Yamaguchi, S. (2007): Seismic interferometric imaging from a point source in the ground, Journal of Seismic Exploration, 15, 323–332. Wapenaar, C.P.A. (2003): Synthesis of an inhomogeneous medium from its acoustic transmission response, Geophysics, 68, 1756–1759. Wapenaar, K. and Fokkema, J. (2006): Green’s function representations for seismic interferometry, Geophysics, 71, SI33–SI46.

894