GB2149503A - Improvements relating to seismic surveying - Google Patents

Abstract

A 3-dimensional seismic survey of subsurface strata produces reflection samples relating to common source-detector midpoints on a rectangular grid in which the lines AB of the grid in at least one direction are in groups in which the lines are more closely spaced than the nearest lines in adjacent groups. The bunching of lines into groups has no effect on the accuracy of the crossline interpolation but enables a group of lines to be surveyed at the same time by a vessel using more than one separate seismic source and/or more than one streamer cable of seismic receivers. It is of particular value when the survey has widely spaced lines so that the surveying of two lines of an evenly spaced grid at once would involve an impractically wide separation of the sources or streamer cables. <IMAGE>

Description

SPECIFICATION Improvements relating to seismic surveying This invention relates to seismic surveying and especially but not exclusively to the so-called three dimensional seismic surveying of subsurface stata under water. It is also applicable to land based surveys although some at least of the advantages obtained from its use in under water surveys would not be obtained.

Seismic surveying under water is of particular importance for underground mineral resources such as, for example, oil and gas. The object of the survey is to detect configurations of subsurface strata which would indicate to an experienced analyst the likelihood of underground reservoirs of oil or gas, say. The surveying technique involves the generation of seismic pulses usually produced by so-called air guns or arrays of such guns, which produce pressure waves in the water which after reflection at the interfaces between the strata are detected by geophones or other seismic detectors, or arrays of such detectors. The timing of the responses picked up by detectors indicates the distance below the surface of the interface where the reflection occurred and from this information a representation of the location of the interface can be produced.

Figure 1 shows a vessel 1 towing a streamer cable 2 and seismic source 4 separately attached to the vessel under the surface of water 3. The cable carries spaced out along its length a plurality of detectors 5A, 5B, 5C, etc. In practice, the seismic source 4 may consist of a plurality of air guns or other seismic pulse generators which are fired simultaneously so as to produce a spherically spreading pressure wave. The disposition of the air guns and the use of arrays of detectors may be chosen to give preference to vertically directed energy.

The detectors 5 are simple transducers which are arranged to pick up pressure waves from the water. For each shot fired, each detector produces a string of responses - commonly of 5 or 6 seconds duration - usually called a seismic trace. In practice, the streamer cable may be several thousands of metres in length with the detectors 5 equally spaced along it at intervals of, for example, 50 metres. The spacing between the detectors and the distance between successive shots fired by the source 4 will depend on the in-line spacing of the subsurface positions where the depths of the stratal interfaces are to be determined. In Figure 1, the bed of the sea or lake on which the vessel 1 is sailing is indicated by the line 6 and two horizontal subsurface stratal interfaces are represented by the lines 7 and 8.The paths of a single seismic pulse from the source 4 are indicated by the lines 9, 10, and 11. The pulse is reflected at the bed 6 and at the interfaces 7 and 8, and the detector 5A picks up the parts of the seismc pulse reflected at a sampling position X. Similarly, the detector 5B picks up the reflections from a sampling position Y, and the detector 5C reflections from a sampling position Z. Clearly, the reflections at the interfaces 7 and 8 will arrive at the detector 5A after the reflection at the bed 6 because of the greater distance that the pulse has to travel.

The timing of the firing of the source 4 may be arranged so that a seismic pulse is emitted at points spaced by half the spacing between the detectors 5 on the streamer cable. As a result of th is, further reflections will be obtained from the bed 6 and the stratal interfaces 7 and 8 at the sampling point X in response to the next pulse emitted by the source 4 after the position shown in Figure 1, and these reflections willl be picked up by the detector 5B. In such an arrangement it will be apparent that there will be obtained from each sampling point X, Y, Z, etc., as many seismic traces as there are detectors on the streamer cable.In order to compensate for the extra distance travelled by the seismic pulses due to the horizontal offset of the source and detector from the subsurface point at which the reflection occurs, what is termed "normal move-out correction" is applied to the reflection data for each subsurface point which enables the "common depth point" traces to be summed or stacked together to improve the signal to noise ratio of the data collected.

For a so-called 3D seismic survey, as shown in Figure 2, the subsurface sampling points are chosen so as to lie on a rectangular, possibly square, grid along parallel lines 12B, 12C ... 12F, which represent the path of the vessel over the sea bed. The sampling points are represented by crosses with the in-line and crossline spacings of the sampling points being 25 metres and 100 metres respectively. The in-line spacing usually lies in the range 6.25 metres to 50 metres, a value of 25 metres being common. The crossline spacing is commonly 75 metres and usually lies in the range 37.5 metres to 100 metres, with up to 200 metres being used sometimes for reconnaissance surveys.In order to obtain a representation of the surfaces followed by the stratal interfaces the analyst has to examine the two-way travel times associated with reflections from each interface at the different sampling points. These travel times will ultimately be converted to depths on the basis of assumed (or measured) speeds of propagation of the pressure waves in the substances involved. Prior to this, the stacked traces will have undergone corrections to allow for the effect of nonhorizontal subsurface stratal interfaces (so-called migration of seismic data). The mathematical techniques involved in these corrections will not be described in detail herein.

It will be appreciated that in order to produce the survey diagrammatically illustrated in Figure 2 using the apparatus described with reference to Figure 1, it is necessary for the vessel 1 to sail along each of the lines ...... 1 2F, which is not only a time consuming operation but is also liable to a certain amount of error because the exact position of the vessel 1 and therefore of the track on which it is sailing depends on the accuracy of its navigational equipment and the skill exercised in keeping the vessel 1 on the desired track.

The survey described above with reference to Figures 1 and 2 is idealised to the extent that it assumes that the streamer cable 2 follows straight behind the vessel 1 along its track, whereas under normal circumstances due to the effects of currents in the water it would be "feathered" to one side or the other of the track. However, this feathering can be monitored on the survey vessel using readings obtained from digital magnetic compasses distributed along the cable.

Figure 3 shows the effect of feathering of the streamer cable 2 and its departure from the track of the vessel 1 represented by the dashed line 1 A. As the detectors 5A, 5B, etc., move further away from the line 1 A so too do the sampled points which are shown as crosses and lie midway between the detectors and the source 4 at the times of firing. The amount of feathering is, of course, exaggerated in Figure 3 for clarity of illustration of the effect.

Recognition of the existence of cable feathering has led to the development of a technique known as "binning" which is used to decide which traces are to be stacked together. A bin corresponds to a rectangular area around a desired sampling point. Given the corrections that can later be applied (the migration mentioned above) the traces will be gathered together for stacking according to the bin in which their source-detector midpoints lie (equivalent to an assumption that the subsurface stratal interfaces are all horizontal). In cases of extreme feathering, certain traces could be omitted altogether from the stacking process.

Figure 4 shows a number of bins along a line of the survey with crosses to represent the positions of the source-detector midpoints for traces that will be stacked together to form the so-called common midpoint stack.

Because the feathering is small relative to the distances the detectors are located behind the source, its main effect is to produce some crossline smearing of the points actually sampled in each bin. Additionally, each line of bins will gather traces predominantly from those shot during one particular traverse of the survey area by the vessel. By monitoring the cable feathering during shooting, the track of the vessel can be offset so as to make the centre of the spread of source-detector midpoints (and not the vessel itself) coincide with the desired line of the survey. This will result in the set of common midpoint stacked traces forming a grid as shown in Figure 2.

Use has been made of dual-line sampling in which the number of lines needed to be sailed is halved by using two seismic sources 13 and 14 as shown in Figure 5, the sources being kept apart by twice the spacing between the lines of subsurfaces sampling points indicated by A and B, and the sources being fired alternately so that the detectors pick up the responses from the sampling points on the lines A and B alternately. With a spacing between the lines A and B of 100 metres, it follows that the sources 13 and 14 would need to be 200 metres apart. A method by which it has been proposed to maintain the sources in their correct positions is to use paravanes, which can successfully be used for separations of the sources 13 and 14 up to 80 metres. It has, however, not been able to be used to keep the sources 200 metres apart with accuracy.

It is an object of the present invention to provide a method by which a 3D seismic survey may be performed with fewer sweeps needed to be sailed by the vessel carrying out the survey, but one in which the accuracy of the survey is not significantly reduced.

According to the present invention there is provided a method of performing a 3-dimensional seismic survey of subsurface strata using seismic source means and a plurality of detectors so disposed along one or more lines relative to the positions at which the source means is fired that the common source-detector midpoints from which the detectors pick up reflections of pressure waves from the source means lie substantially on a rectangular grid, wherein the lines of common midpoints on the grid in at least one direction are in groups of the same size within which the lines are relatively closely spaced and the spacings between the nearest lines of adjacent groups are more than the spacing between the lines within a group.

The lines of a group are preferably equally spaced and the spacings between the nearest lines of adjacent groups are equal, thereby to simplify the mathematical operations needed to effect the interpolation transverse to the lines. The groups may each include two lines.

In carrying out a survey of strata under water, there are three possibilities for the arrangement of the seismic source means and the seismic detectors towed in the water by vessel. Firstly, the seismic detectors may be on a single streamer cable and the seismic source means may comprise two or more separate sources spaced apart laterally by twice the distance between the lines of a group and fired cyclically.

Secondly, the seismic source means may be a single source and the seismic detector may be on two or more streamer cables spaced apart laterally by twice the distance between the lines of the group. Thirdly, there may be several sources and several streamer cables spaced apart laterally and towed by more than one vessel if necessary so that a group of lines of the survey can be recorded by each vessel sailing a single line.

Where several separate sources are provided they must be fired one at a time in cyclic succession with sufficient time interval between the firings for the reflections from the greatest depth of interest of one firing to be received before the next firing is made. This means that the common midpoints on one line of a group will be staggered relative to those on another line of the group. Whilst the effects of this staggering can be compensated by in-line interpolation, it can also be avoided by trailing the sources at different distances behind the vessel so that any errors which might arise from the in-line interpolation are not incurred.

According to one aspect of the present invention there is provided a method of carrying out a 3-dimensional seismic survey of subsurface strata under water using a vessel on the water towing seismic source means and a streamer cable having a plurality of seismic detectors in the water, seismic pulses from the source means being picked up by the detectors after reflection from the interfaces of the strata, the vessel performing a plurality of sweeps along parallel lines so as to obtain at discrete positions on a 2-dimensional grid the responses from the detectors representing the depths of the interfaces of the strata, from which displays representing surfaces followed by the stratal interfaces can be derived by interpolation in the 2 dimensions, wherein the seismic source means includes two or more separate sources which are fired cyclically with a sufficient interval between pulses for the detectors to pick up the useful reflections of each pulse before the next occurs, and which are spaced apart laterally by their attachments to the vessel by a significant distance but less than twice the distance between adjacent parallel lines of the survey.

The invention enables two or more lines of reflection samples to be obtained from a single sweep of the vessel, so that both the time and the cost of a survey can be reduced. The survey resulting from the use of the invention has unequally spaced lines of reflection samples, but the unequal spacing does not have a significant effect on the accuracy of the definition by the samples of the profiles representing the stratal interfaces transverse to the lines. It can be shown, for example, that a survey with a crossline structure having samples at spacings of 25 metres and 175 metres alternately defines the stratal interfaces with practically the same accuracy as one with a crossline spacing constant at 100 metres. Therefore if a crossline spacing of 100 metres is adequate for a survey then one having spacings of 25 metres and 175 metres alternately is also adequate.The seismic sources may be trailed from booms on the vessel or paravanes may be used. The lateral spacing between the lines of samples produced during the same sweep of the vessel can be maintained with greater accuracy than the lateral spacing between lines of samples produced during different sweeps of the vessel. The technique of the invention is termed herein paired-line or bunched-line sampling.

Although the advantage obtained in an under-water survey wherein a single pass of the vessel can produce two lines of the survey cannot be obtained with a land based survey, the paired-line or bunched-line technique of the invention can also be used on land and may provide an advantage in certain kinds of terrain where a regular spacing of the lines is awkward to use.

According to a second aspect of the present invention there is provided a method of carrying out a 3-dimensional seismic survey of subsurface strata using seismic source means and a plurality of detectors so disposed along one or more lines relative to the positions at which the source means is fired that the common source-detector midpoints from which the detectors pick up reflections of pressure waves from the source means lie substantially on a rectangular grid, wherein the lines of common midpoints on the grid in at least one direction are in equally spaced pairs with the spacings between corresponding lines of adjacent pairs being equal and more than twice the spacing of the lines within a pair.

According to a further aspect of the present invention there is provided a method of carrying out a 3-dimensional seismic survey of subsurface strata under water in which seismic source means is towed in the water along a preselected path with a streamer cable having a plurality of seismic detectors in the water being towed along a similar or adjacent path, seismic pulses from the source means being picked up by the detectors after reflection from the interfaces between the strata, the preselected path including a plurality of parallel lines so as to provide at discrete positions on a 2-dimensional grid responses from the detectors representing the two-way travel times of reflections from the interfaces, from which displays representing surfaces followed by the strata interfaces can be derived by interpolation in 2-dimensions, wherein the seismic detectors are carried on two or more streamer cables, the streamer cables being spaced apart laterally by a significant distance but less than the distance between adjacent parallel lines of the survey. The use of two streamer cables enables two or more lines of reflection samples to be obtained at a time, the lines of samples being closer together than half the distance between the lines of the survey.

Preferably the seismic detectors are located side by side on the two or more streamer cables and the seismic source means is fired at such positions along the parallel lines that the sampled points at which the reflections occur, which reflections are picked up by the detectors, lie substantially on straight lines transverse to the parallel lines.

The seismic source means may comprise one or more separate sources towed by the same vessel as is towing the streamer cables, towed by one or more other vessels on a parallel track or towed by the same vessel and one or more other vessels on a parallel track. Booms or paravanes may be used to hold the streamer cables at the required lateral spacing and if required similar apparatus may be employed to keep a source on a desired track.

In order that the invention may be fully understood and readily carried into effect it will now be described in more detail with reference to accompanying drawings, of which Figures 1, 2, 3,4 and 5 relate to previously proposed methods of seismic surveying and have already been referred to; Figure 6 is a plan view illustrating a vessel performing a survey in accordance with an example of the present invention; Figure 7 shows one possible distribution of sampled points of a survey carried out in accordance with the invention; Figure 8 is a diagram illustrating the subsurface structural sampling transverse to the lines of the survey of Figure 7; Figure 9 shows a modification of Figure 6; Figure 10 is a diagram of an alternative form of survey according to the invention;; Figure ii illustrates transverse sampling applied to the survey of Figure 9; Figure 12 is a diagram of a vessel performing a survey in accordance with another embodiment of the present invention; Figure 13 is an idealised diagram of the survey points produced by the embodiment shown in Figure 12 showing a crossline interpolation; Figures 14, 15 and 16 are diagrams of arrangements of one or two vessels with sources and streamer cables for performing surveys; Figure 17 is a diagram showing an effect of feathering on surveyed points in one type of survey; Figure 18 is a diagram showing the surveyed points of another type of survey; and Figure 19 shows in idealised form the points surveyed by a single streamer cable subject to feathering.

In Figure 6, the vessel 1 is towing a streamer cable 2 in the same way as the vessel shown in Figure 1 already referred to. Detectors 5A, 5B, 5C and 5D are shown on the cable 2 and are representative of a much larger number of detectors distributed along a streamer cable which may be up to, for example, 2.5 kilometres in length. In addition, the vessel 1 is towing two seismic sources 20 and 21 which are held out on opposite sides of the cable 2 by paravanes with a spacing in between the sources of, for example, 50 metres.

The sources 20 and 21 are fired alternately with sufficient time between firings that the reflected seismic pulses will have been received from the greatest depth of interest in response to one pulse before the next pulse is emitted. The source-detector midpoints - and thus the theoretical subsurface sampling points - due to pulses from source 20 are represented by the crosses 22A, 22B, 22C and 22D. Similarly, those due to the seismic pulses from source 21 are represented by the crosses 23A, 23B, 23C and 23D. Because of the alternate firing of the sources 20 and 21, the points 22 and 23 are staggered relative to one another along the survey lines.By virtue of the lateral displacement of the two sources and careful steering of the vessel in response to information on the extent of the cable feathering, traces can be gathered into bins and stacked together to produce traces with enhanced signal-to-noise characteristics lying on, or close to, the survey lines A & B. It is desirable to arrange the distance between the consecutive firings of source 20 (and similarly source 21) to be half the distance between adjacent detectors so that each common midpoint bin gathers traces from the same detectors.If the traces gathered together in a line of bins were to consist alternately of those from even numbered detectors and those from odd numbered detectors, the "centres of gravity" of the midpoints in the bins (the stacked trace positions) would zig-zag when cable feathering was present and the consistent distance between the stacked traces associated with lines A and B would be lost.

Figure 7 illustrates in idealised form a survey which could be produced by the apparatus shown in Figure 6 in which the vessel 1 sails along lines 200 metres apart and produces the two lines A and B of sampled points 25 metres apart. The lines A and B are 25 metres apart because the sources 20 and 21 are 50 metres apart.

Within the binning process the set of lines A and the set of lines B should be treated quite separately as this will result in a consistent separation of the lines of stacked traces within each pair as long as the separation of sources 20 and 21 is consistent. This result is independent both of the extent of cable feathering (which changes only slowly relative to the time interval between successive shots) and of the presence of errors in the steering of the vessel. In this example there is 100 metres between detectors along the streamer cable and 25 metres between a firing of source 20 and that of source 21 (and vice versa). This gives a stacked trace spacing of 50 metres in each sub-survey staggered by 25 metres relative to each other.

The traces gathered within each bin are summed together (stacked) so as to reduce the effects of noise.

This stacking process is preceded by the correction of each trace to allow for the horizontal distance covered, this correction being referred to as "normal moveout correction". Migration of the stacked traces can then be effected by processing in known manner after a rectangular grid of traces has been obtained using mathematical techniques to be described shortly.

The recorded seismic traces are usually stored in digital form and the normal moveout corrections, other pre-stack processes and the stacking itself performed by computing apparatus which may be carried by vessel 1 but is, more usually, remote from it. In Figure 7, the returns are arranged so that the lines travelled by the vessel 1 are parallel to the X direction and the Y direction is at right angles to the lines. The T direction relates to the two-way travel times of the pressure waves from the sources which are representative of the depths of the points at which the reflections occur. In the analysis of the returns interpolation along the lines is first carried out so that the effects of the staggering of the sampled points can be compensated.

Interpolation is then carried out in the Y direction transverse to the lines, and one example of such interpolation is shown in Figure 8.

In Figure 8, the travel time T of corresponding reflection points on lines A and B is used to provide values for the interpolation to produce the line 25 which joins up the corresponding reflection points at all of the sampled points across the survey in the particular transverse line concerned. The dotted values C represent those of points on lines C which with the lines A would form the lines of a conventional regularly sampled survey. These regular samples would be the usual output from the interpolation in the Y direction.

Although the interpolation could be performed in they domain, where y is the direction transverse to the lines of the survey, it is more efficient in the use of computing facilities to perform the interpolation calculations in the frequency (s) domain where functions in s are related to those in y by Fourier transforms.

Suppose that the line 25 of Figure 8 is a function T = f(y), then the corresponding Fourier transform may be termed F(s). It should be noted that, in the mathematics that follows, the function f(y) is one which can be adequately defined by regular samples at intervals of 0.5 in y (this is equivalent to saying that F(s) is zero outside the range -1 < s < +1), the inter-pair distance is 1 and the intra-pair distance is a. If only the samples along the lines A had been taken corresponding to the use of a single seismic source only, then these samples could be represented as the function TTTf(y) and the Fourier transform ofTTT f(y) is TTT f(y) where TTTf(y) = TTT(y).f(y) and

3 (y) being the Dirac delta function.

Since the samples obtained along the lines B are not used in TTT f(y), it follows that the function f(y) is sampled only half as frequently as is required. In order to remedy this situation, TTTf(y+a) is also measured corresponding to the samples along the lines B where 'a' is possibly much smaller than unity (unity being the spacing of the lines A in Figure 8). The function f(y+a) has the Fourier transform e2waisF(s). The Fourier transform ofTTTf(y+a) isTTTf(y+a).

In the s-domain with -1 < s < 1,wehave TTTf(y) = F(s+1) + F(s) + F(s-1) and TTTf(y+a) = e2wai(stl)F(s+1) + e2"ais.F(s) + e2sai(s-1) F(s-1) For s > 0, F(s+1) = 0. We can therefore eliminate F(s-1) from the above expressions and rearrange the terms to give

For s < 0, we have F(s-1) = 0. Again, eliminating F(s+ 1) from the above expressions and rearranging the terms, we have

Since F(s) can be defined for s < 0 and s > 0 (and as TTT f(y) j s=0 when s=0), it follows that using the expressions derived above we can reconstruct f(y) using Fourier transforms from TTT f(y), and TTT f(y+a), provided that 'a' is not an integer.The effect of 'a' being an integer would of course be that f(y) had only been sampled at integer values ofywhich we know to be inadequate.

Similarly, it can be shown that the functions f(y), f(y+a) and f(y+2a) can be used at a spacing of 1.5, or f(y), f(y+a), f(y+2a) and f(y+3a) at a spacing of 2.0, and so on.

Figure 9 shows a modification to Figure 6 in which one ofthe seismic sources 25 is carried by a paravane, and the other source 26 is towed behind a paravane by a length of cable 27. The length of the cable 27 is arranged to be equal to twice the distance sailed by the vessel 1 between a firing of the source 25 and the next firing of the source 26. The advantage of this arrangement is that the common midpoints are aligned across the line and not staggered as in Figure 6, so that there is no need for in-line interpolation prior to crossline interpolation.A disadvantage of the arrangement is that the lateral spacing between the sources may be less accurately controlled than in Figure 6 particularly in the presence of crossline currents, although measures can be taken to improve the accuracy, for example, by making the paravane and source configurations symmetrical as far as the drag due to the water is concerned and correcting the lateral separation for the effects of crossline currents using a controllable paravane. The technique can be used with more than two sources, and the means for locating the sources need not be as shown in the Figure provided that the relative positioning of the two sources is maintained.

Figure 10 shows a distribution of sampled points of a survey carried out using a vessel with three spaced seismic sources 40m apart which are fired in cyclic succession as the vessel progresses along the line of the survey. In the example shots are fired every 25m and the detectors are spaced at 150 metre intervals along the cable. A cable with detectors spaced 50 metres could be employed. The additional traces produced by each shot would be used to help the process of noise attenuation (by known frequency domain filtering methods) prior to the stacking of the traces from every third detector. Stacking the traces from all the detectors would yield individual lines (so-called 2D seismic survey lines) with a subsurface sampling interval of 25 metres.In the processing, in-line interpolation would be used initially to provide transversely aligned values for use in the reconstruction of regular crossline sampling. As it proceeds along its course, the vessel surveys three parallel lines, A, B and C. As a number of lines is surveyed at the same time, a difficulty could be experienced in obtaining adequate in-line sampling because of the need to allow sufficient time between firings for the most remote returns to have been received by the detectors. This difficulty could be overcome by arranging for the vessel to travel sufficiently slowly along its course, subject to the limitation that if the vessel goes too slowly control of the cable will be lost. With 75 metre in-line sampling and the use of frequency domain interpolation it is necessary to use termporal band limiting of the data to prevent aliasing.

Figure 11 shows an example of a profile obtained in the crossline direction by means of a survey carried out as shown in Figure 10. The sample values for the lines A, B and C are shown as solid lines and for comparison equally spaced values A' and A" for use with those on lines A are shown as dotted lines.

The theory of reconstruction from interlaced samples can be applied in a similar way to that described for paired samples to this sampling in y, the crossline coordinate.

Assume that, based upon the crossline structure present and the proposed termporal band limiting of the data, the wavefield at the surface produced by reflection from the subsurface strata can be adequately sampled at 100 metre intervals. The survey as described above with reference to Figures 6,7 and 8 is carried out with pairs of lines with 25 metres between the lines within a pair and 200 metres between the pairs. Each pair of lines is shot during one traverse by the vessel using alternate starboard and port seismic source firing. Reconstruction of the surfaces at 100 metre sampling is then performed using the above mathematics and fast Fourier transforms. An algorithm for producing the transform may be found in "Fundamentals of Geophysical Data Processing" by J. F. Claerbout, McGraw-Hill.

Compared with a conventional 100 metre 3D survey by the use of the invention the costs of obtaining the samples have been approximately halved whereas the data processing costs have only increased by a very small amount because of the unevenly spaced samples.

In theory, the survey could have been carried out with bunches of three lines using three separated sources, with the vessel sailing along lines 300 metres apart. Certain practical difficulties which might arise with such a survey are that obtaining adequate in-line sampling might be difficult because three lines are being sampled at the same time, and the survey would be more susceptible to the effect of independent errors between lines within the bunch than is the paired sampling. However, such a survey could be performed bearing in mind the above difficulties with every attempt being made to minimise these errors.

In the example of paired-line survey described above the starboard and port sources would have been held 50 metres apart. This would be achieved using paravanes on either side of the vessel; booms could be used for small spacings, e.g. 25 metres. It is important to keep the distance between the sources consistent and both booms and paravanes do this adequately. It will be evident that the invention is of greatest value when crossline sampling in excess of 50 metres is all that is required. A paired-line or bunched-line survey can always be turned into a regular 3D survey by shooting additional lines of samples should the need arise to confirm the results obtained by using more frequent sampling to increase the resolution.

The spacing of the sources is chosen to be as large as possible consistent with the need for the spacing to be controlled as accurately as possible, beaing in mind that certain types of errors in the reconstruction process are increased as the separation of the sources becomes smaller. To some extent the greater accuracy with which a smaller spacing can be held - particularly if booms can be used in place of paravanes offsets the increase in the errors as a result of the use of a smaller spacing.

One way in which the survey data obtained by the use of the invention could be processed is to stack separately the samples derived from the two (or more) sources, to give two or more sub-surveys. Suppose there are two subsurveys. The best place in the processing to combine the two sub-surveys into one is immediately after stacking rather than following in-line migration which might introduce errors that would degrade the performance of the reconstruction process. The cheapest way - if frequency domain migration is to be used - would be to perform the crossline migration after the reconstruction without leaving the frequency domain and thereafter do the in-line migration.Temporal band-limiting, to ensure that the functions defining the surfaces are only aliased within each sub-survey to the extent that the reconstruction process can handle, must be performed prior to the Fourier transforms.

The only increase in data processing costs beyond those associated with the equivalent, conventionally shot, 3D survey arises from the reconstruction itself. Since the work in the transform (frequency) domain consists only of some multiplication of complex numbers, almost its entire cost comes from the (fast) Fourier transforms.

No special considerations are required when performing the 3D migration once the regular grid of stacked traces has been constructed.

From practical experience with the invention it has been found that it works well on correctly sampled data, that reconstruction errors only occur on the traces between adjacent input sample pairs, that the process is more sensitive to errors as the ratio of the inter- to intra- pair separation increases (as theoretical studies would lead one to expect) and that temporal band-limiting of the input data can indeed control the extent of the spatial aliasing and thence the performance of the reconstruction algorithm.

During model studies in which the only errors in the data samples were due to mispositioning of the survey vessel, samples reconstructed from a paired-line 3D survey in which there was a consistent separation between the two sources were found to be less subject to error than those collected in a regular 3D survey (with comparable steering errors) once the magnitude of these steering errors passed a certain threshold. In practice, this threshold would always be exceeded.

Figure 12 shows another method of performing a survey according to the invention in which the vessel 1 tows two streamer cables 30 and 31 held 50 metres apart across the track of the vessel by paravanes 32 and 33. A single seismic source 34 is towed by the vessel 1. Seismic receivers on the streamer cables 30 and 31 pick up the reflections of pulses emitted by the source 34 from the common midpoints represented by crosses in the Figure. There are two lines of common midpoints separated by 25 metres, half the distance between the streamer cables. The receivers on the two streamer cables are 50 metres apart along the cables and corresponding receivers are preferably at equal distances behind the vessel 1 so that the lines joining the pairs of common midpoints are substantially at right angles to the track of the vessel 1.

Figure 13 shows three pairs of lines of sampled midpoints 35,36 and 37, the pairs being at 200 metre intervals and the lines of each pair being 25 metres apart. The line 38 is an example of a crossline profile produced by interpolation from six sample values 39 to 44; the values could be the two-way transit time of a vibration from the source to a stratal interface and back to a receiver at different positions on the same interface and with allowances made for normal moveout correction and other sources of inaccuracy.

In contrast to the method of seismic surveying described above in which paired line or bunched line 3-dimensional surveys are performed using a plurality of separate seismic sources fired in cyclic sequence with a single streamer cable, the use of two or more streamer cables means that the detectors on the cables receive reflected energy from sampled points aligned across the track of the vessel. This has the advantage that interpolation along the track of the vessel is not needed before interpolation transverse to the track can be performed, and therefore inaccuracies due to the along-track interpolation are avoided.On the other hand, the spacing between the streamer cables can vary slightly due to differences between the "feathering" of the cables, which will influence the accuracy of the interpolation transverse to the track; however, the accuracy of the spacing between streamer cables towed by the same vessel, and therefore between the lines of sampled points, is much higher than that resulting from separate sweeps of the vessel along parallel lines using a single seismic source and a single streamer cable. Another advantage arises from the use of a single source in that it can be fired twice as frequently as one of a pair of sources which have to be fired alternately; this gives rise to a higher in-line sampling density.

Figures 14, 15 and 16 show other arrangements of two streamer cables with one or more sources. In Figure 14 there is an additional source near the start of one of the streamer cables. In Figure 15 a second vessel is towing a second source. In Figure 16 the second vessel is towing the only source. Many other configurations would, of course, be possible with, for example, three or possibly four streamer cables. A difficulty which would be encountered using many sources is that they must be fired serially with sufficient interval between firings to allow for the longest expected two-way travel time for the seismic vibration; this can restrict the minimum in-line spacing of sampling, It would be possible for each streamer cable to respond to only its own source attached to it so that the lines of midpoints surveyed are separated by the distance between the cables.

Figures 17 and 18 show another advantage of using a single source with two streamer cables over using a single streamer cable and two sources. They show the effect of feathering of the cable(s) on the centres of gravity of the source/receiver midpoints of a 2n trace/n fold survey, which points should ideally be coincident and lie on the lines shown. In both Figures the centres of gravity, shown as crosses, should lie on the dotted lines but alternate ones in each line are just above the line. In Figure 17, which shows the case for two streamer cables and one source, the vibration in position of the crosses is in phase in the two lines, so that the lateral spacing of the centres of gravity is substantially constant.In Figure 18, which shows the case for one streamer cable and two sources, the variation in position of the crosses is out of phase in the two lines, so that lateral spacing of the centres of gravity changes by twice the displacement due to the feathering. As explained above, since the lines of samples are relatively close together it is important that the spacing be maintained at the required value for accuracy in the crossline interpolation, and clearly the use of two streamer cables and a single source is preferable in this respect.

Figure 19 shows how feathering causes the lateral offset of the centre of gravity and why it differs with alternate sampling positions. The lines A to F represent six consecutive lines of source/receiver common midpoints with greatly exaggerated though linear feathering of a streamer cable. With a 2n trace/n fold survey the in-line distance between sampling positions (i.e. the distance between receivers) between firings, as a result of which the reflection samples relating to one sampled point are received by odd-numbered receivers and the reflection samples relating to the next sampled point are received by even-numbered receivers. This is represented by the oblongs G and H in the Figure, from which it can be seen that the centres of gravity are biased alternately to one side than the other as a result of the feathering.

Claims (10)

1. A method of performing a 3-dimensional seismic survey of subsurface strata using seismic source means and a plurality of detectors so disposed along one or more lines relative to the positions at which the source means is fired that the common source-detector midpoints from which the detectors pick up reflections of pressure waves from the source means lie substantially on a rectangular grid, wherein the lines of common midpoints on the grid in at least one direction are in groups of the same size within which the lines are relatively closely spaced and the spacings between the nearest lines of adjacent groups are more than the spacing between the lines within a group.

2. A method according to claim 1 wherein the lines within a group are equally spaced and the spacings between the nearest lines of adjacent groups are equal.

3. A method according to claim 1 or 2 wherein each group includes two lines.

4. A method according to claim 1,2 or 3 for performing a survey under water using a vessel on the water towing the seismic source means and the seismic detectors in the water, wherein the seismic detectors are on a single streamer cable and the seismic source means comprises two or more equally spaced separate sources equal in number to the lines of a group, spaced apart laterally by twice the spacing between the lines of a group and fired cyclically, whereby a group of lines of the survey can be recorded whilst the vessel is sailing along a single line.

5. A method according to claim 1, 2 or 3 for performing a survey under water using a vessel on the water towing the seismic source means and the seismic detectors in the water, wherein the seismic source means is a single source and the seismic detectors are on two or more streamer cables equal in number to the lines of a group and spaced apart laterally by twice the spacing between the lines of a group, whereby a group of lines of the survey can be recorded whilst the vessel is sailing along a single line.

6. A method according to claim 1,2 or 3 for performing a survey under water using one or more vessels on the water towing seismic source means and the seismic detectors in the water, wherein the seismic source means comprises two or more laterally spaced separate sources and the seismic detectors are on two or more laterally spaced streamer cables, whereby a group of lines of the survey can be recorded whilst the or each vessel is sailing along a single line.

7. A method according to claim 4 or claim 6 wherein the longitudinal separation of the separate sources and the relationship between their times of firing are such that the common midpoints on at least some of the lines of a group are aligned substantially at right angles to the lines of the group.

8. A method of carrying out a 3-dimensional seismic survey of subsurface strata under water using a vessel on the water towing seismic source means and a streamer cable having a plurality of seismic detectors in the water, seismic pulses from the source means being picked up by the detectors after reflection from the interfaces of the strata, the vessel performing a plurality of sweeps along parallel lines so as to obtain at discrete positions on a 2-dimensional grid the responses from the detectors representing the depths of the interfaces of the strata, from which displays representing surfaces followed by the stratal interfaces can be derived by interpolation in the 2 dimensions, wherein the seismic source means includes two or more separate sources which are fired cyclically with a sufficient interval between pulses for the detectors to pick up the useful reflection of each pulse before the next occurs, and which are spaced apart laterally by their attachment to the vessel by a significant distance but less than twice the distance between adjacent parallel lines of the survey.

9. A method of carrying out a 3-dimensional seismic survey of subsurface strata using seismic source means and a plurality of detectors so disposed along one or more lines relative to the positions at which the source means is fired that the common source-detector midpoints from which the detectors pick up reflections of pressure waves from the source means lie substantially on a rectangular grid, wherein the lines of common midpoints on the grid in at least one direction are in equally spaced pairs with the spacings between corresponding lines of adjacent pairs being equal and more than twice the spacing of the lines within a pair.

10. A method of carrying out a 3-dimensional seismic survey of subsurface strata under water in which seismic source means is towed in the water along a preselected path with a streamer cable having a plurality of seismic detectors in the water being towed along a similar or adjacent path, seismic pulses from the source means being picked up by the detectors after reflection from the interfaces between the strata, the preselected path including a plurality of parallel lines so as to provide at discrete positions on a 2-dimensional grid responses from the detectors representing the two-way travel times of reflection from the interfaces, from which displays representing surfaces followed by the stratal interfaces can be derived by interpolation in 2-dimensions, wherein the seismic detectors are carried on two or more streamer cables, the streamer cables being spaced apart laterally by a significant distance but less than the distance between adjacent parallel lines of the survey.

11 f A method of performing a 3-dimensional seismic survey of subsurface strata according to claim 1 substantially as described herein with reference to the accompanying drawings.