FR2764993A1 - METHOD FOR DEVELOPING IMPROVED TIME OR DEPTH STRUCTURE CARDS FROM TWO-DIMENSIONAL SEISMIC DATA - Google Patents

Abstract

The invention concerns a method for producing improved maps of time or depth structure based on two-dimensional seismic data, consisting in migrating non-migrated 2D data at a velocity function V1 to produce an initial 3D map for a given reflector R2. The method is characterised in that it consists in de-migrating all the points P1 of the initial map to obtain de-migrated points P2, selecting the points P2 which are located on or in the proximity of acquisition lines L1.....Ln, migrating the selected points P2 at a velocity function V1 to obtain migrated points P3, reading the point P4 of the reflector R2 nearest to the point P3, de-migrating each point P4 to obtain a point P5 to which is assigned a horizontal component of the orthogonal slope associated with point P2, migrating the points P5 to obtain an intermediate 3D map and proceeding by iterative method until points P3 points P4 are convergent.

Description

Method for developing improved time structure maps or
depth from two-dimensional seismic data
The present invention relates to a method for developing improved maps of time or depth structure from two-dimensional seismic data.

 To analyze an environment using seismic data recorded along acquisition lines, the interpreters prefer to interpret migrated seismic data rather than non-migrated data, because the migrated sections reflect more the image of the geology of the middle basement and because said migrated seismic sections have better lateral resolution.

 The techniques for acquiring 2D data and making the corresponding migrated sections are well known to specialists and will therefore not be described, the migrated sections being obtainable after a depth migration or a time migration, it being specified that it is possible to pass from 'a time migration to a deep migration. The time or depth migration procedures have the same properties as regards their seismic correspondences. A depth migration requires knowing the interval velocities while a time migration accommodates a simpler definition of the field of migration speeds. .

 When we interpret migrated sections obtained from 2D acquisition lines that intersect, we are confronted with the problem linked to crossing faults between acquisition lines. Crossover faults can originate from the processing and / or interpretation of seismic data, or even tectonics from the environment, especially if we are dealing with complex tectonics.

 To construct a map of a given horizon, (map which is considered to be a 3D map), the time migrated sections are used in the usual way. In the event of a crossing fault, either a reinterpretation or an average of the differences is carried out and if necessary a kriging is carried out.

 This procedure has at least two drawbacks.

The first lies in the fact that we are trying to solve the problem of crossing faults without looking for their origin. The second resides in the fact that the 3D map is not calibrated (put-tie in English) with the initial 2D data because it is made the assumption that the seismic waves remain permanently in the plane passing through each of the lines d 'seismic acquisition, which is the case only if the horizons detected are planar and horizontal and if there are no lateral variations in speed, or if the structures are cylindrical and the line perpendicular to a generator.

An improvement to the method considered as classic and described succinctly above, was proposed by DN WITHCOMBE and
RJ CARROLL. This improvement is described in an article published in
GEOPHYSICS, vol. 59, No. 8 (July 1994), pages 1121-1132.

 For a given reflector, the authors propose to demigrate the 3D map associated with said reflector, using a constant demigration speed or a speed grid obtained from migration velocities or summation speeds (stack velocities in English) along 2D lines. For each point of the 3D map, we calculate the spatial and temporal displacement vectors (difference between the migrated point and the demigrated point), then we apply to the migrated points 2D initial time of each acquisition line, the vector component displacement perpendicular to the line. In this way, we get new points which are used to build a new 3D map. The process is repeated until the differences between migrated points are minimal from one 3D map to another.

The main disadvantages of the WITHCOMBE method and
CARROLL resides in the fact that the demigration speed of each 3D map is locally constant and in the fact that the time migration speed used to carry out the 2D time migrated sections and the demigrated component parallel to the 2D line are not used to correct the initial 3D map. Under these conditions, there is no correspondence between the demigrated points from various 3D maps and the initial time migrated data.

Another method proposed by Richard T. Houck et al., "The
Leading Edge ", 1996, v. 15, nr. 8, concerns the development of 3D maps from 2D seismic data.

 As in the case of the WHITCOMBE-CARROLL method, the initial data are the points made on the migrated 2D sections. It is assumed that the 2D migrations are exact and therefore that the migrated reflectors are correctly displaced in the plane of the line. We then calculate the lateral displacement, due to the lateral dip, to find the exact position of the reflector. The method proposes to calculate all the possible displacements for each point. An elimination operation is planned in the case where the possible displacements of different points overlap. At the end of this procedure we obtain what the authors call a constrained surface (the points on this surface represent all the possible dips which may have been encountered by the 2D lines).

Using this surface, we can generate a set of 3D maps.

 A disadvantage of the HOUCK method is that it uses as a basic assumption that the seismic processing is exact and that the 2D migration has correctly positioned the data in the plane of the line, the basic assumption never being questioned throughout the treatment.

 The object of the present invention is to propose a method capable of allowing the development of improved 3D maps or so-called true 3D maps, and this, from 2D data.

 The present invention will be described in detail for a time migration but it should be noted that it can be implemented directly for a deep migration.

 Indeed, there is no fundamental difference between a time migration and a depth migration. Time migration, as mentioned above, is only slightly less sensitive to the velocity field than depth migration. A demigration which is an inverse operation to migration in the time domain, is equivalent to a "modeling" in the depth domain.

The subject of the invention is a method of the type consisting in a) using seismic data acquired along acquisition lines
L1 .... Ln and which, after possible preliminary treatment, are considered
as being 2D seismic data, b) carrying out a first 2D migration of said 2D seismic data
with a first law of speeds V1, c) point reflectors of interest R1 ..... Rn on the migrated sections as well
obtained, d) construct at least for a reflector of interest R2 a first card
3D by interpolation of the 2D migrated points, e) migrate said first 3D map with a second law of speeds
V2, so that at each migrated point P1 of said first 3D map
corresponds to a demigrated point P2, and it is characterized in that it also consists in f) selecting from the demigrated points P2 those which are on or at
proximity of each acquisition line L1 ... Ln, g) carry out a second 2D migration of said demigrated points P2 with
the same speed law V1 used in the first 2D migration, from
so as to obtain new 2D migrated P3 points, h) search for each new 2D migrated P3 point, the most
near P4 located on said reflector of interest, i) note the distance separating each new point P3 migrated 2D from the point
the closest corresponding P4, j) carry for each nearest point P4 reflectors
R1 .... Rn a second demigration with said first law of
speeds V1, so that at each nearest point P4 corresponds
a new demigrated point Pg, k) assign to each new demigrated point P5 the component of the dip
orthogonal to the line associated with said demigrated point P2, 1) construct, by migration of points Pg, a 3D intermediate map,
then m) iterate steps e) to l) for the construction of other
successive 3D intermediate maps until convergence between the
new migrated points P3 and the closest points P4.

 An advantage of the present invention lies in the fact that a real 3D map is obtained from 2D data both in the time domain and in the depth domain. The depth migration of the reflector is obtained by working directly in depth, as will be described briefly later, or by following the last 3D time migration with a demigration and a 3D ray tracing.

 Finally, another advantage of the present method is that it can make it possible to highlight pointing errors by comparing the demigrated data of the intersecting acquisition lines.

Other advantages and characteristics will appear better on reading the description of the method according to the invention given, by way of example, in the time domain, as well as the appended drawings in which
FIG. 1 is a partial and schematic representation of a 3D cube migrated over time,
FIG. 2 is a schematic representation of a first 3D map obtained from the time migrated dots of the 3D cube of FIG. 1 which relate to the reflector R2,
FIG. 3 is a view of the plane X, Y representing three acquisition lines L1 to L3 with the associated bands b1 to b3,
- Figure 4 is a partial and schematic view of a time migrated section X, TM of the 3D cube of Figure 1 and on which a single time migrated reflector R2 is shown.

 - Figures 5 to 7 are synthetic representations respectively of an initial 3D map, an intermediate 3D map and a final 3D map.

2D data are recorded beforehand, along acquisition lines or directions, five of which are shown in FIG. 1 and referenced L1 to L5. The acquisition lines L1 to L5 intersect and form any grid which can, for example, be oriented along the axes
X and Y, the axis TM being the migrated time axis which, in the depth domain, would be the depth axis.

 The preliminary seismic processing consists in carrying out a sum domain (stack in English), in the usual way and it is the data of the sum domain which are two-dimensional or 2D and which are used in the next step of time migration. When we migrate time 2D data relating to line L2 for example, with a speed law V1, we obtain a section migrated time S, that is to say a plane passing through line L2, and on which we points to the different reflectors detected during the reception-recording emission of the waves emitted in the medium under consideration.

For clarity, we will refer in the following to the single line
L2, it being specified that the method takes into account all the acquisition lines. Likewise, only three reflectors R1 to R3 are represented diagrammatically in the migrated time section S and for the description only reference is made to reflector R2.

 Another step consists in defining around each acquisition line L1 to L5 a narrow band b1 to bg. Advantageously, the width of this strip is of the order of the intertrace, that is to say for example around 25 m. Of course, this bandwidth can be less than 25 m and moreover it will be reduced, preferably, as and when iterations which will be discussed later.

 A first 3D map (initial map) relating to the reflector R2 is constructed by interpolation of the 2D dots migrated time, corresponding to the reflector R2. Only a contour C of such a 3D time map is shown very schematically in FIG. 2. The process for developing said first 3D map is well known to specialists and will therefore not be described in detail. The choice essentially depends on the operator. For example, we can interpolate the 2D migration speeds to obtain a speed law for the construction of the initial 3D map.

 In another step, a first demigration is carried out by migrating each of the points P1 of the first card C with a second law of speeds V2, the choice of which essentially depends on the operator. For example, we can interpolate 2D migration speeds. Each point P1 corresponds to a demigrated point P2.

 Among all the demigrated points P2, one selects for the acquisition line L2 considered, those which fall on the line L2 or which are located in the corresponding band b2 surrounding L2. Then we proceed to a second time migration of the points P2 with the speed law V1. For each point P2, as regards the reflector R2 considered, there corresponds a point migrated time P3. This is what is shown diagrammatically in FIG. 4 which is an X-TM plane in which the reflector R2 is represented.

 Another step consists, for each point P3, in finding the point P4 closest to P3 and located on the reflector R2. The distance from point P3 to point P4 is noted and stored. We then proceed to a second 2D time demigration with the speed law V1, of all the points P4. Each point P4 corresponds to a demigrated point P5 in the surface plane X, Y containing the line L2 and which should not be too far from the first demigrated point P2. The component of the dip orthogonal to the line L2 associated with the point P2 is assigned to each point P5.

 In another step, a second intermediate 3D map relating to the reflector R2 is constructed by migration and from all the points P5 of the surface X, Y of the block.

 The demigration-migration steps are then carried out in the manner indicated above, but with the 3D intermediate map obtained previously.

 Successive iterations are carried out until points P3 and points P4 converge, that is to say until the difference A between points P3 and P4 is as small as possible. In this way we obtain a final 3D map.

 Figures 5 to 7 show an example of an initial synthetic map (Figure 5) and 3D intermediate maps (Figure 6) and 3D final maps (Figure 7).

 As stated previously, the method of the present invention can be implemented directly in the depth domain.

Indeed, one can proceed in the following manner which is strictly equivalent to that described in the time domain.

 In a first step, we migrate in depth the 2D stack data not migrated with a speed law V1. Then, we point out the horizons of interest in deep migration. After which and by interpolating migrated 2D depth data, a first 3D map or initial map is produced. We then carry out a modeling of the points P1 of the initial 3D map by means of a ray tracing to obtain the demigrated points P2. We select the points P2 which are inside narrow bands as indicated in the example migration time. The selected points P2 are migrated depth with the same speed law V1 as that used for the previous depth migration or first migration, so that points p3 are obtained. For each point P3, we search for the nearest point P4 located on the reflector and we model the points P4 ,, by ray tracing, in the plane of the corresponding line, so as to obtain points P5, then we assign to each point P5, the horizontal component of the dip, orthogonal to the line associated with point P2 ,. We migrate in depth 3D the points P; so as to obtain an intermediate 3D depth map and then proceeds to the iterations indicated above with the intermediate 3D depth maps, to obtain the final 3D depth map.

 However, it is also possible to produce a depth map from the final time migrated map obtained according to the method of the invention. For this, we perform a 3D demigration, similar to the 3D demigration which is carried out during the iterative process, so as to pass to the stack domain. Each demigrated point in the stack area is passed in depth using ray tracing. For this, from the surface (X stack, Y stack, given by demigration) we launch a ray. The initial direction of the radius is provided by demigration and it is subsequently changed according to the contrasts of the impedance of the velocity model that is used. We continue this process until the journey time becomes equal to the stack time (demigrated) for the point in question and we repeat the process for all points. The result is a scatter plot. The surface that passes through this point cloud is the depth migrated reflector.

 When there are interval velocities, the method can be used to create the reflectors of a velocity model from the 2D line plotting.

 This is the case if you have a 2D seismic campaign or extract a few 2D lines from a 3D campaign to quickly have a first model without using all the data (no 3D seismic migration for example).

If these lines are pointed in depth (after migration seismic depth before or after stack), one can either work in depth as described in the method according to the invention, or perform 2D modeling to obtain the section with zero offset (zero-offset ), then the 2D time migration of each line and use the present method in the time domain. The 2D velocity model used for the time migration can be deduced from the previous depth model by calculating the velocities
RMS (quadratic speeds).

Claims (7)

1. Method for developing an improved map from 2D two-dimensional seismic data, of the type consisting of a) using seismic data acquired along acquisition lines  L1 Ln and which, after possible preliminary treatment, are considered  as being 2D seismic data, b) carrying out a first 2D migration of said 2D seismic data  with a first speed law V1, c) point the reflectors of interest R1 Rn to the migrated sections as well  obtained, d) construct at least for a reflector of interest R2 a first card  3D by interpolation of the 2D migrated points, e) migrate said first 3D map with a second law of speeds  V2, so that at each migrated point P1 of said first 3D map  corresponds to a demigrated point P2, characterized in that it also consists in f) selecting from the demigrated points P2 those which are on or at  proximity of each L1 acquisition line. lez g) proceed to a second 2D migration of said demigrated points P2 with  the same speed law V1 used in the first 2D migration, from  so as to obtain new 2D migrated P3 points, h) search for each new 2D migrated P3 point, the most  near P4 located on said reflector of interest, i) note the distance separating each new point P3 migrated 2D from the point  the closest corresponding P4, j) perform for each nearest point P4 reflectors R1 .. Rn  a second time demigration with said first speed law  V1, so that at each nearest point P4 corresponds a  new demigrated point P5, k) assign to each new demigrated point P5 the component of the dip  orthogonal to the line associated with said demigrated point P2, l) construct, by migration of points Ps, a 3D intermediate map,  then to m) iterate steps e) to 1) for the construction of other  successive 3D intermediate maps until convergence between the  new migrated points P3 and the closest points P4. 2. Method according to claim 1, characterized in that it is determined around each acquisition line L1 ... Ln a narrow band bl ... b n, and in that the selected demigrator points P2 are in said bowl bands. Well 3. Method according to claim 2, characterized in that the width of each strip is equal to about 25 m. 4. Method according to claim 1, characterized in that the steps of migration and demigration are carried out in the time domain, so as to obtain a final 3D map migrated time. 5. Method according to claim 1, characterized in that the steps of migration and demigration are carried out in the depth domain, so as to obtain a final migrated depth 3D map. 6. Method according to claim 5, characterized in that the deep demigration is a modeling by ray tracing. 7. Method according to claim 4, characterized in that the final 3D migrated time map is demigrated time and in that at each demigrated time point the ray tracing process is applied until the travel time becomes equal at the time migrated.