|
|
 |
|||||||||||||||||
|
|||||||||||||||||
 |
|||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
SPECIAL SECTION: Hydrogeophysics |
Geological Survey of Canada
Corresponding author: apugin{at}NRCan-RNCan.gc.ca
Multicomponent seismic reflection methods are a new tool for oil and gas exploration and reservoir monitoring (Miles 1988), but such technologies have not yet been extensively exploited for near-surface exploration related to hydrogeological and/or geotechnical investigations. With the advantage of relatively inexpensive recording systems for near-surface applications, we show that the use of multicomponent high-resolution seismic reflection methods has great potential as a new means of observing and characterizing the physical parameters of the shallow subsurface, and in particular of groundwater reservoirs.
Over the last decade, there has been a marked increase in the use of shear-wave reflection profiling in near-surface investigations. This has been facilitated in part by the development and increasing acceptance of landstreamers (Inazaki, 2004; Pugin et al., 2004). To date, almost all shallow seismic shear-wave reflection work has considered only the SH-mode, using source and receivers which are oriented perpendicular to the survey line. Recent tests (Pugin et al., 2008) have highlighted the potential of SV data recording using a vertical component geophone. We present results of further experimentation which involves the observation and analyses of P, SV, and SH body waves from multicomponent (two- or three-component) records obtained with high-frequency vibroseis equipment with an orientable source (vibrating in the vertical or horizontal direction), a broad-band sweep (10–350 Hz), and a short, 0.75-m spaced receiver array mounted on sleds (landstreamer).
We first examine SV sections obtained over buried valleys in two very different geological settings. In eastern Canada, we collected seismic reflection data over a valley in Precambrian bedrock buried beneath thick Holocene marine silts/clays. In contrast, we also present data from western Canada, where thick glacial deposits including several coarse-grained units, infill an overdeepened valley in central British Columbia. In both cases, we found that all components were present in the recorded data, to some degree, irrespective of the source orientation. To further investigate and better understand this, we have conducted some detailed nine-component experiments to examine all reflection phases on three-component (3-C) records using vertical, inline-shear, and transverse-shear source orientations. These results have led to some interesting observations of distinct differences in the dominant frequency between SH and SV reflection energy. Finally, we examine SV wave polarization and the possible improvements in reflection profiles that can be obtained by stacking data that have been rotated to the optimum elliptical polarization direction. The aim of the work is ultimately to better understand and resolve the subsurface architecture and stratigraphy, and to provide improved and more reliable data to groundwater modelers.
Seismic reflection 3-C data acquisition
The Geological Survey of Canada has recently developed a 48-channel, 3-C landstreamer and tested it over buried bedrock valleys infilled with various subsurface lithologies. The landstreamer (Figure 1) is built with 3-kg metal sleds connected using low-stretch rope. The number of receivers and the receiver spacing can vary depending on the near-surface velocities and the targeted depths of observation. We typically use 24–72 receiver sleds at spacings ranging from 0.75 m to 3 m. On each sled, we attached a 3-C geophone unit constructed inhouse using 30-Hz omnidirectional geophone elements oriented in three directions, one vertical and two horizontal, inline and crossline. The source is an IVI minivib minibuggy with all the recording capabilities installed within the cab of the vehicle. The minivib can be operated with the vibrating mass oriented vertically or in any horizontal direction. The source spacing is measured using a system mounted on the minivib itself; it is preset to the desired shot spacing which typically ranges from 1.5–6 m.
|
A buried bedrock valley beneath thick marine clays in the St. Lawrence lowlands near Ottawa yields high quality SV stacked sections using this acquisition system (Figure 2). This area was occupied by the Champlain Sea 10,000 years ago as the Laurentide ice sheet was retreating. The homogenous and well stratified sediment deposited in the Champlain Sea transmits very high-frequency energy over relatively long distances (Hunter et al., 1984), and it is an ideal medium for recognition and separation of various types of body waves (P, PS, and S). This SV section is representative of the very high quality of the reflection data that can be acquired in this geological setting. These data were recorded with a receiver spacing of 0.75 m, a shot spacing of 3 m, an inline horizontal source, and inline and vertical receivers. Shear-wave energy was observed over a frequency band from 
50 Hz to 300 Hz. Data processing included a phase rotation, bandpass filtering, gain adjustments, NMO corrections, and stacking. Further discussion on polarization and phase rotation will be given throughout the text.
|
The example presented in Figure 2 was obtained in a near-ideal geological setting for shallow seismic reflection methods, but it is well known that a low water table in coarse gravel or till is a more challenging environment to image using P-wave seismic reflection techniques. This was the situation in a recent survey in the Kelowna area in south-central British Columbia, where the goal was to delineate the stratigraphy of deep glacial valleys infilled with up to several hundred meters of glacial and glaciomarine deposits. This survey was initially planned using the classic P-wave approach of a vertical source and vertical receivers. However, as we started surveying in this area, we found that the vertical mode of the source did not produce useable P-wave reflection energy, but using the H1 component and vibrating horizontally inline gave better results. Figure 3 shows example sections from this survey where very little P-wave energy was recorded on the vertical receiver component. These data were recorded with a receiver spacing of 1.5 m and a shot spacing of 3 m. The right side of the P-wave section shows little more than stacked background noise; the left side shows some P-wave energy, but complex lateral velocity changes in the near-surface sediments create an almost unsolvable static problem. In contrast, when we examine the SV data, the result is a good quality stacked section. NMO shear-wave velocities ranged from 250 m/s to 450 m/s at 250 m depth. In a raw 2-C record (Figure 4), we can see little or no P-wave energy present in the vertical component; instead the record is dominated by surface waves and shear-wave phases. This is not simply a function of the source orientation; similar records were obtained during tests with the source vibrating in vertical mode. As will be shown, P-wave energy can be recorded using either vertical or horizontal vibrating source modes.
|
|
Examining complete 9-C data
The two examples above highlight some of the recent success we have had in using the inline shear component (SV) for shallow seismic reflection profiling. The data presented above led us to question the assumption that shear-wave reflection surveys should consider only the SH component to avoid interference with P-waves and/or converted waves and to question whether we really understood how any body-wave reflections (P, SH, and SV) were produced and transmitted through the ground.
To answer these new questions, we conducted a 2D, 9-C reflection profiling test over an edge of a buried valley filled with marine clay sediments. The nine-components were created by running the line three times using three different source orientations (V = vertical, H1 = inline horizontal, and H2 = transverse or crossline horizontal), each time recording the data from all three receiver orientations. The 47-channel data were recorded with a receiver spacing of 0.75 m and a source spacing of 3 m. Nine raw records were produced at each shot location (Figure 5). As mentioned above, the Champlain Sea sediments provide a near ideal medium for shallow seismic reflections, and we can clearly observe P, PS, and SH/SV reflections on these records; surface-wave interference is minimal and the signal/noise ratio is very high. In Figure 5, we have highlighted three important observations:
|
|
Using this same 2D, 9-C data set we have produced a 250-m long section using a simple processing sequence that includes a large gain-balancing AGC window, band-pass filter, NMO correction, and CMP stack. The nine S-wave sections corresponding to the three source and three receiver orientations (upper three rows of Figure 6) all show a good quality bedrock reflection across most of the section at 0.25 s two-way traveltime (TWTT) or 20 m depth.
At this site, the SV section produced using the horizontal inline source and vertical receivers (S(H1,V) in Figure 6) produced the highest-resolution stacked section. This SV section has a frequency content that ranges from 50 Hz up to 300 Hz with ultrashallow reflections (from 2 m depth), and has the highest amplitude and most coherent reflections observed within the mud succession. The SV stacked section obtained using a vertical source and a vertical receiver (S(V,V) in Figure 6) is also of very high quality.
As expected, the best SH reflection is obtained using a crossline source and crossline receivers (S(H2,H2, in Figure 6). However, as discussed above and highlighted by the red circles (Figure 5), the frequency content is significantly lower (below 100 Hz) than observed on the SV sections. We show that a P-wave section can be processed using the vertical component even when the source is vibrating horizontally (bottom row of three sections Figure 6). The inline P(H1,V) section shows the signal with the highest frequency content and most structural detail from the top of the bedrock (reflection at 0.03 s TWTT).
Horizontal source orientation
We previously pointed out a frequency difference observed in the shear-wave bedrock reflection event depending on the receiver orientation (red circles, Figure 5). In this case, the H2 orientation of the source produces a relatively low-frequency reflection event on the H2 receiver, a slightly higher-frequency event on the H1 (inline) receiver, and the highest-frequency event on the vertical receiver. To further understand and quantify this observation, we conducted an experiment in which we left the receiver array in one position, and recorded data with the vibrator in horizontal mode (20–350 Hz sweeps), rotating the mass in increments of 15o from a crossline (H2) position at 0o up to the reverse crossline position at 180o passing by the in-line (H1) position at 90o. We then analyzed the bedrock reflection event and produced spectral energy plots of the signal frequency versus the angle of the source (Figure 7).
|
On the vertical receiver, we observe exclusively the high-frequency SV-wave phase with the same amplitude trend as described for the H1 records. These data help characterize the frequency differentiation between SH and SV reflections (Pugin et al., 2008). To our knowledge, similar observations have not been made in deeper oil exploration seismic surveys. However, in such surveys, horizontal vibrators are usually vibrating below 40 Hz, and over this more limited frequency range, it may be impossible to observe frequency differentiation.
Filtering, polarization, and rotation
With the acquisition of multicomponent data, it is possible to consider whether further improvements in the quality of the final processed section can be achieved using polarization analyses and rotation to enhance the signal-noise ratio of the shallow seismic reflection data.
The first critical step in processing a high-resolution multicomponent data set is to precisely determine the frequency band of the signal. Based on our experience, it is not difficult to remove surface waves with a band-pass filter; in cases where the frequencies of surface waves and shear-wave reflections are too close, a carefully applied f-k filtering may be more effective. As we have seen, SH or SV phases may be characterized by distinctly different frequency bands, and an appropriate band-limiting filter should be applied before performing any polarization analyses.
Our observations show that in general, the optimum elliptic polarization for SV energy is seen in the (H1,V) plane (Figure 8). The other planes (H2,V and H2,H1) show more complex ground motions. Based on these observations, we chose to concentrate on the (H1,V) plane. To get an optimum stacked reflection section, we consider the polarization of the wave as an ellipsoid; the rotation of this ellipsoid (Figure 9) provides a means to separate the data into particle motions along the long (a) and the short (b) axes of the polarization. Using a sliding window, we calculate the eigenvector representing the long axis of the polarized ellipsoid (Figure 10). The rotated traces show the enhancement of the reflection signals in the new reference system which is better aligned with the actual ground motions.
|
|
|
|
In acquiring and analyzing 3-C shallow reflection data, we have made some initial, tentative steps toward a better understanding and appreciation of the potential of multicomponent shallow seismic reflection surveying. It is becoming clear that there are many ways to produce different reflection phases in the subsurface. We have observed that vibrating the ground in essentially any direction yields a full range of P-wave and S-wave energy in conducive settings.
The usual assumption that SH data are produced using a crossline source and acquired with crossline receivers or that SV data require an inline source and inline receiver must be reassessed. While it appears that differentiation of SH- and SV-data wave is most effective relative to the receiver orientation, in general it appears that any source direction produces both SH and SV energy. Typical vertical sources used for traditional P-wave reflection surveys likely produce ample SV and SH reflections—though we usually do not look for them or do not record for long enough to observe them. P-wave reflections are usually recorded only on vertical receivers, but in our tests they were produced regardless of source orientation.
These observations lead to the conclusion that multicomponent seismic reflection data are easy to produce. They can also be easily recorded using a landstreamer array fitted with 3-C geophones. The landstreamer removes the need to pick up and move geophones and geophone cables along a line, and greatly improves the efficiency of recording and orienting a large number of recording channels while decreasing the risk of errors in recording the geometry.
In our test area, we have observed a differentiation in frequency between SH and SV phases, with the SV phase being characterized by significantly higher frequencies and, therefore, yielding higher-resolution reflection sections. The frequency differentiation between SH and SV waves has to be further tested in other geological settings, and particularly in less layered media, to determine the effect of anisotropy resulting from the layering within the subsurface. We have also observed significant interference between reflected and refracted phases at very high acoustic-impedance interfaces. A better understanding of the refracted phase may be of great use for understanding the nature of the bedrock-sediment interface. Finally, we have made some first attempts to improve the final stacked reflection section by rotating multicomponent data to a variable reference system which aligns with the actual ground motion.
The complexity and the richness of shallow multicomponent seismic reflection data needs to be further investigated with much more detailed experimentation and modelling. These data hold the key to being able to consider the anisotropy of stratified sediments and the lateral and vertical changes in physical properties that are necessary to improve comprehension and characterization of groundwater resources and flow.
Suggested reading
"Evaluation of anisotropy by shear-wave splitting" by Crampin (GEOPHYSICS, 1985). "Shallow seismic reflection mapping of the overburden-bedrock interface with the engineering seismograph—Some simple techniques" by Hunter et al. (GEOPHYSICS, 1984)."High resolution reflection surveying at paved areas using S-wave type land streamer" by Inazaki (Exploration Geophysics, 2004). "Three-component seismic data: Researcher's toy or interpreter's tool?" by Miles (TLE, 1988). "Near-surface mapping using SH-wave and P-wave seismic land-streamer data acquisition in Illinois, U.S." by Pugin (TLE, 2004). "SV-wave and P-wave high resolution seismic reflection using vertical impacting and vibrating sources" by Pugin et al. (SAGEEP Proceedings, 2008). "Shear waves from 3D 9-C seismic reflection data: Have we been looking for signal in all the wrong places?" by Simmons and Backus (TLE, 2001).
Acknowledgments:
We thank Timothy Cartwright, Marten Douma, and Robert Burns for their support in the field and for the designing of equipment. A special acknowledgment to Steve Sargent, Illinois State Geological Survey, for creating all sorts of very useful electronic devices. We warmly thank Rob Huggins from Geometrics, Carleton University (Ottawa, Ontario) and Golder Associates, (Toronto, Ontario) for their generous support during this research.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
