SEISMIC REFLECTION The seismic reflection method absorbs more than 90% of the money spent world-wide on applied

geophysics. Most surveys are aimed at defining oilbearing structures at depths of thousands of metres using hundreds or even thousands of detectors. However, some reflection work is done by small field crews probing to depths of, at most, a few hundred metres. The instruments used in these surveys were originally very simple but may now have as much in-built processing power as the massive processing laboratories of 20 years ago. Field operators need to have some understanding of the theory behind the options available. 12.1 Reflection Theory Ray-path diagrams, as used in Chapter 11, provide useful insights into the timing of reflection events but give no indication of amplitudes. 12.1.1 Reflection coefficients and acoustic impedances The acoustic impedance of a rock, usually denoted by I, is equal to its density multiplied by the seismic P-wave velocity. If a seismic wavefront strikes a planar interface between two rock layers with impedances I1 and I2 at right angles (normal incidence), the amplitude of the reflected wave, as a percentage of the amplitude of the incident wave (the reflection coefficient, RC) is given by: RC = (I2 I1)/(I2 + I1) If I1 is greater than I2, the coefficient is negative and the wave is reflected with phase reversed, i.e. a negative pulse will be returned where a positive pulse was transmitted and vice versa. The amount of energy reflected first decreases and then increases as the angle of incidence increases. If the velocity is greater in the second medium than in the first, there is ultimately total reflection and no transmitted wave (Section 11.1.5). However, most small-scale surveys use waves reflected at nearly normal incidence. 12.1.2 Normal moveout The true normal-incidence ray cannot be used in survey work, since a detector at the shot point would probably be damaged and would certainly be set into such violent oscillation that the whole record would be unusable. Geophones

Figure 12.1 Derivation of the normal moveout equation for a horizontal reflector. To is the normal incidence time.

are therefore offset from sources and geometric corrections must be made to travel times. Figure 12.1 shows reflection from a horizontal interface, depth d, to a geophone at a distance x from the source. The exact hyperbolic equation linking the travel time T and the normal incidence time To is established by application of the Pythagoras theorem. For small offsets, the exact equation can be replaced by the parabolic approximation, which gives the normal moveout (NMO), T To, directly as a function of velocity, reflection time and offset. T To = x2/2V 2To Since V usually increases with depth and To always does, NMO decreases (i.e. NMO curves flatten) with depth. Curved alignments of reflection events can be seen on many multi-channel records (Figure 12.2). Curvature is the most reliable way of distinguishing shallow reflections from refractions. 12.1.3 Dix velocity If there are several different layers above a reflector, the NMO equation will give the root-mean-square (RMS) velocity defined as: V RMS 2 = (V1 2 t1 + V 22 t2 + Vn2 tn)/Tn where tn is the transit time through the nth layer, velocity Vn, and Tn is the total transit time to the base of the nth layer. Interval velocities can be calculated from RMS velocities using the Dix formula: V DIX 2= (Vn1 2Tn1 Vn 2 Tn)/(Tn1 Tn)

Figure 12.2 Enhancement seismograph record showing curved alignment of reflections (thick line). The earlier events were produced by refractions. Note that on Channels 11 and 12 the strong refracted wave completely overwrites the reflection. The variable area presentation used is popular for reflection work since it emphasizes trace-to-trace correlations, although some information is lost where traces overlap. The subscripts n 1 and n denote, respectively, the top and bottom of the nth layer. RMS velocities are normally slightly higher than true average

velocities, since squaring the high velocities increases their influence on the average. Significant errors can arise if RMS velocities are used directly to make depth estimates but these are generally less than the errors introduced by the use of the NMO equation to estimate velocity using reflections from interfaces that may well not be horizontal. Dix conversion may not help very much in these cases. 12.1.4 Effect of dip If the source is placed at the centre of the geophone spread, the curves obtained over horizontal interfaces will be symmetrical about the source point. If, however, the reflector has a uniform dip , the reduction in travel path on the updip side of the shot compensates to some extent for the offset, and some travel times will be less than the normal-incidence time (Figure 12.3). The minimum time 2d cos()/V is recorded at a distance 2d sin() from the shot, on the updip side. The reflected ray rises vertically to this point, about which the moveout curve is symmetrical. Dip effects in shallow reflection surveys are detectable only for very large dips or very long spreads.

Figure 12.3 Effect of dip on a single-fold record. Rays are reflected from the dipping interface as if derived from the image point S_ at depth 2d cos below the surface, where d is the perpendicular distance from the shotpoint to the interface. The normal incidence travel time is 2d/V but the shortest travel time is for the ray which is vertical after reflection. An identical moveout hyperbola would be produced by a shot at point G and a horizontal interface at depth d cos .

Figure 12.4 Paths for multiple reflections. (a) Simple multiple. (b) Peg-leg. (c) Intra-formational multiple. 12.1.5 Multiple reflections A wave reflected upwards with high amplitude from a subsurface interface

can be reflected down again from the ground surface and then back from the same interface. This is a simple multiple. Two strong reflectors can generate peg-leg and intraformational multiples (Figure 12.4). Multiples are difficult to identify with certainty on single traces. They can sometimes be recognized on multi-channel records because they have moveouts appropriate to shallow reflectors and simple time relationships with their primaries. 12.2 Reflection Surveys Reflected waves are never first arrivals, so clear-cut reflection events are seldom seen. Oil-industry techniques for improving signal-to-noise ratios can be used for shallow work and simple versions of the programs used are incorporated in the software supplied with the latest generation of 12- and 24-channel seismographs. 12.2.1 Spread lengths The distance from the source to the nearest geophone in a shallow reflection survey is usually dictated by the strength of the source (and the need to protect the geophone) and may be as little as 2 m when a hammer is being used. Even with explosives or heavy weight drops, minimum offsets of more than about 10 m are unusual when observing shallow reflections. A reflection spread can be much shorter than a refraction spread used to probe to similar depths, but with powerful sources and multi-channel recording, the furthest geophone may be more than 100 m from the source. The optimum spread length can be determined only by experiment, since the most important factors are the arrival times of the noise trains associated with the direct wave and any strong refracted waves. Field work should begin with tests specifically designed to examine these arrivals, generally by using elongated spreads. 12.2.2 Arrays Ideally, reflected energy should arrive after the near-surface waves (groundroll and refractions) have passed but this may not be possible if the depth of investigation is very small. In such cases, geophones may be connected in arrays to each recording channel. Reflected waves, which travel almost vertically, will reach all the geophones in an array almost simultaneously but the direct waves will arrive at different times and produce signals that can interfere destructively. The efficiency with which a wave is attenuated by an array is defined by its relative effect (RE) compared to the effect of the same number of geophones placed together at the array centre. The variation of the RE with apparent wavelength (which for the direct wave is equal to the true wavelength), for a linear array of five geophones equally spaced on a line directed towards the shot point, is shown in Figure 12.5. Non-linear arrays produce more complex curves. Simple arrays are preferred in the field, since mistakes are easily made in setting out complicated ones. The range of frequencies over which attenuation of the direct wave occurs is proportional to array length and it may be

necessary to overlap the geophones in adjacent arrays. It would be unusual in a shallow survey to use more than five geophones per array.

Figure 12.5 Relative effect (RE) of an array of five equispaced in-line geophones. The 100% level would be attained with zero spacing between the geophones. The apparent wavelength is equal to the actual wavelength divided by the sine of the angle between the wavefront and the ground surface, and is infinite for a wave rising vertically and equal to the true wavelength for the direct wave. Attenuation is concentrated between values of apparent wavelength divided by geophone spacing of about 1.2 and 7. With 2 m spacing, a 500 ms1 wave would be attenuated at frequencies of between about 35 and 200 Hz. 12.2.3 Shot arrays Seismic cables for use with only 12 or 24 channels are not designed with arrays in mind, and non-standard connectors may have to be fabricated to link the geophones to each other and to the cable. It may be easier to use arrays of shots instead. A shot array using explosives usually involves simultaneous detonation of charges laid out in a pattern resembling that of a conventional geophone array. If an impact source is used with an enhancement instrument, the same effect can be obtained by adding together results obtained with the impact at different points. This is the simplest way of reducing the effects of surface waves when using a hammer. 12.2.4 Common mid-point shooting Improving signal-to-noise ratios by adding together several traces (stacking) is fundamental to deep reflection surveys. In shallow surveys this technique is normally used only to stack (enhance) results obtained with identical source and detector positions. If, however, the data are recorded digitally, NMO corrections can be made (although not in the field) to traces produced with different sourcereceiver combinations. The technique normally used is to

collect together a number of traces that have the same mid-point between source and receiver (common midpoint or CMP traces), apply the corrections and then stack. The number of traces gathered together in a CMP stack defines the fold of coverage. Three traces forming a single synthetic zero-offset trace constitute a 3-fold stack and are said to provide 300% cover. The maximum fold obtainable, unless the shot point and geophone line are moved together by fractions of a geophone interval, is equal to half the number of data channels. Figure 12.6 shows the successive geophone and source positions when a six-channel instrument is used to obtain 300% cover. Special cables and switching circuits are available for use in deep reflection surveys, but CMP fieldwork with the instruments used for shallow surveys is very slow and laborious. The need to combine traces from several different shots makes it difficult to do CMP processing in the field.

Figure 12.6 CMP schematic, for 3-fold cover with a 6-channel system. Shot points A, B, C and D are progressively one geophone group interval further to the right. Note that the distance between reflection points (depth points) on the interface is only half that between geophone groups on the surface. Shots A and D have no depth points in common.

Figure 12.7 Effect of dip in CMP shooting. In contrast to single-fold shooting (Figure 12.3), the shot points as well as the geophone locations are different for the different traces. Shot points and detector locations are equivalent and the depth point on the reflector moves up dip as the offset increases. The moveout equation is most easily derived by noting that the path from source to detector is equal in length to the path SG from the source to the detector image point, and that the geometric relationships between similar triangles imply the equality of all the lengths marked y. The Pythagoras relationship can be applied to the triangle SGP, and the times can be obtained by dividing the distances by V . Thus, To = 2d/V and T = SG/V . The geometry of a CMP shoot differs from that for single-fold coverage, and the effect of dip is therefore different (Figure 12.7). If the interface dips at an angle , the velocity deduced from a CMP stack is equal to V /cos and the depth is equal to the length of the normal incidence ray from the common mid-point to the interface. In contrast to the single-fold gather of Section 12.1.4, the minimum time is associated with the normal incidence ray. The aim of stacking is to produce a noise-reduced seismic trace that approximates to the normal incidence trace, i.e. to the trace that would have been produced had the source and detector been coincident. The initials CMP replaced an earlier acronym, CDP (common depth point) used for the same method. Referring to depth points (reflection points) as common implies that all the reflections in a gather have come from the same point on the subsurface interface, which is true only for horizontal interfaces. 12.2.5 Depth conversion Reflection events are recorded not in depth but in two-way time (TWT). Velocities are needed to convert times into depths, but the Dix velocities (Section 12.1.3) obtained from NMO curves may be 1020% in error, even

Figure 12.8 Geometric distortion on seismic sections. The image is of a small graben structure beneath an unconformity. The position of the true fault plane BB (indicated by the dashed line) can be estimated from the positions of the terminations of the sub-horizontal reflectors representing the sediment fill within the graben (although care must be exercised because many of the deeper sub-horizontal events are multiples). The event AA is the seismic image of BB. It is displaced because the techniques used to display the data assume that reflections are generated from points vertically beneath the surface points, whereas they are actually generated by normal-incidence rays that are inclined to the vertical if reflected from dipping interfaces (Section 10.3.2). The reflections from the fault and the opposite side of the graben cross over near the lower symbol A, forming a bow-tie. Convex-upward reflections near point C are diffraction patterns generated by faulting. for horizontal reflectors. Interpretations should be calibrated against borehole data wherever possible, and field crews should always be on the lookout for opportunities to measure vertical velocities directly. 12.2.6 Geometric distortion Seismic reflection data are normally presented as sections prepared by playing out, next to each other and vertically down the sheet of paper, the traces from adjacent CMP gathers. Such sections are subject to geometric distortion. Artefacts such as displaced reflectors, diffraction patterns and bow-ties, described in Section 10.3.2 as affecting radar sections, also appear on seismic imagery, as is shown in Figure 12.8.