AU2005242215A1

AU2005242215A1 - Wide Aperture Seismic Interpretation

Info

Publication number: AU2005242215A1
Application number: AU2005242215A
Authority: AU
Inventors: James Howard Leven
Original assignee: JAMES LEVEN
Current assignee: Leven James Howard Dr
Priority date: 2005-12-12
Filing date: 2005-12-12
Publication date: 2007-06-28

Description

Patent application for WIDE APERTURE SEISMIC INTERPRETATION SUBMISSION DETAILS Title Wide aperture seismic (WAS) interpretation Inventor Dr James Howard Leven GeoSeis Pty Ltd, Perth, Australia.

Postal Address: PO Box 250 Street Address: 12 Chelmsford Ro Mount Lawley WA 6929 ad Mount Lawley WA 6050 IPAustralia Customer ID: 6810152654 W AS i.H.Leven i Page I FIELD OF THE INVENTION

U

d The present invention discloses a method of interpreting wide aperture seismic data. It is applicable in the oil and gas exploration and production industry.

C"l I/3 BACKGROUND OF THE INVENTION C"l Description of Related Art Processing of seismic data for the exploration and exploitation of subsurface resources c has traditionally utilized the technique of stacking to increase signal-to-noise ratio and to elucidate the subsurface structure. The observed seismic data record of the wavefield either on the land surface (onshore) or in near-surface water (offshore), from a source of seismic energy (typically either explosive shot or vibratory source, or airgun), can be processed by grouping traces according to some criteria, such as common mid-point between the source and receiver locations, and stacking the data in these common groups after applying a correction for differing offsets between the source and receiver.

However, such an approach to the processing and interpretation of seismic data makes the following simplifying assumptions 1. The geological structure of the subsurface is essentially horizontally layered; 2. The mid-point of the source and receiver location and the travel time uniquely defines the reflection point in the sub-surface, independent of offset; 3. Refraction of the seismic energy does not significantly alter the path of the seismic energy from a straight line between the source or receiver point and the reflection point.

Whereas corrections can be applied within the processing stream to address some of these shortcomings, these corrections introduce further assumptions, which also incur limitations in the accuracy and resolution of the processed data. The assumptions inherent in seismic processing which involves the gathering and stacking of the seismic data will be violated in any area of geological complexity.

Furthermore, to avoid the deterioration of the contributions to the stack due to differential stretching of the normal move-out correction at long offsets, data are usually muted to remove long-offset and refracted arrivals. It is, however, these longer offset data which can provide the constraint for the seismic wavespeeds necessary for depth W AS Sdetermination.

a) It is an object of the present invention to provide a method of processing wide aperture 0 seismic data which ameliorate the shortcomings of these prior art techniques involving c1 the gathering of seismic data. The present technique utilizes seismic wavefield recorded over a wide aperture of offsets between the seismic source and the receiver, and efficiently utilizes both the near-source and long-offset seismic data to generate a depth 4T model that accurately represents the structure and seismic wavespeed of the subsurface.

In WAS SUMMARY OF THE INVENTION CN The present invention provides a method of processing and interpreting seismic data a, which utilities seismic ray theory to build a depth model of the sub-surface geology. The Q model represents the interface geometry between the various geological strata or units c1 and seismic wavespeed of these geological strata or units.

The method utilizes wide aperture seismic data recorded as individual common shot or kn common receiver gathers, and iteratively constructs the model by propagating rays with NS a range of takeoff angles into the model from the source or receiver location appropriate

C

for each gather. The travel times and amplitudes of rays which return to the recording C< interface are used to construct a travel-time distance curve.

In 0 Rays can be modelled to either reflect or refract at each interface, and to be transmitted or reflected as either a compressional or shear wave within each layer.

Description The spirit and scope of the present invention is not restricted to the described embodiments, as illustrated below, but encompasses any invention disclosed herein, explicitly or implicitly, and any generalization thereof.

To illustrate the method, a simple two-layer marine model (Figure 1) has been generated, in which compressional rays are propagated at a range of take-off angles from a source at position of X=1000 m and a depth of Z=10 m. This figure displays the rays (drawn in red) which reflect at the horizontal interface at a depth of 3000 m, and return to the seafloor at the depth of 1000 m, to be recorded by an ocean bottom seismic array.

Figure 2 illustrates the rays propagating in the same model which refract in the uppermost sub-sea layer, and are turned by the wavespeed gradient in that layer. These rays which bottom at depths between 1000 m and 3000 metres also return to the seafloor, and are denoted diving rays. In this simple model, the diving rays turning in this layer would be recorded at the seafloor in the offset range of 1.5 to 10.5 km.

Recording the arrival times of this diving ray group in this offset range, and from a variety of source points, provides constraint on the wavespeed in this layer.

Arrival times and offsets of rays having a common propagation path define a travel-time distance curve or trajectory for that ray group. Figure 3 plots the travel times and amplitudes of rays propagating as a diving ray group or a reflected ray group, showing the two ray groups illustrated in Figures 1 and 2. The arrival times are plotted as red dots, and the corresponding relative energy of the arrivals are plotted as blue dots. The travel times have been plotted with a linear move-out applied (reduced travel time plot).

X. JJ. :X.4-CUT: M-T :~illli: In this illustrative model, the reflected rays have a distinct increase in relative energy near a distance of 6500 m (offset 5.5 km), with relatively low amplitudes in the near-

U

O offset region and considerably higher amplitudes of the post-critical reflections at offsets exceeding 5.5 km.

To illustrate the application of the method, Figure 4 illustrates a model denoted the Marmousi-2 model (Martin, Marfurt, and Larsen, SEG Abs, 2004) which has greater 1fl structural complexity than the model presented in Figure 1, and incorporates a range of c features typical of hydrocarbon exploration areas. Martin (2004, MSc thesis, University Cl of Houston) has calculated synthetic seismograms, using the finite difference c computation method, for a range of source and receiver geometries. Figure 4 plots the Sinterfaces of this model, with a portion of a specified interface the left-hand segment 0 of interface 1034 highlighted in red.

Figure 5 plots the ray paths (in red) propagating within this model, with an illustrative range of take-off angles, emanating from a source at X= 4475 m. Figure 5 shows rays in this ray group reflecting at the 1034 interface, and returning to the seafloor to be recorded by an ocean bottom seismic array.

Figure 6 plots the calculated travel time distance curve or trajectory of the ray group in the Marmousi-2 model, which has a single reflection at the interface 1034, and which has a compression downgoing and upcoming ray type, with the ray emanating from a source at X=4475 m, and as recorded by the ocean bottom recording array. This figure plots the travel times in red dots as standard unreduced two way travel times versus distance. The corresponding relative energy of the ray group are plotted as blue dots, which illustrate the low amplitudes of this ray group at near offsets and the relatively higher amplitudes recorded at offsets greater than 3 km.

Figure 7 plots this travel times of this ray group using reduced travel-times, applying a linear move-out correction with a reducing speed (denoted Vr) of 3000 m/s. Plotting the travel time curve using a reduced travel time can more readily illustrate the effect of structure, in this case of interface 1034, on the resultant travel times. The travel time trajectory is a piecewise continuous function of offset and travel time.

The present invention entails calculating a statistical metric, denoted the statistical coherence functional, which measures the degree of coherence of the recorded wavefield along the travel time distance trajectory of the various ray groups. The preferred implementation of the method uses the seismic semblance as the realization of this statistical coherence functional, as demonstrated below.

X. JJ. 4-CUT: M

J

Figure 8 plots the travel time distance curve (in black) of the ray group, having a single reflection at H034, onto the finite difference synthetic seismogram (in blue) of an ocean

U

a bottom array calculated for a source point at X=4475 m in the Marmousi-2 model (Martin, op. cit.).

Figure 9 illustrates the same travel time distance curve (in black), and data (in blue), but plotted using a reduced travel time and a reducing speed of 3000 m's. Figure 10 zooms tt in, showing multiples and other events, often with far greater amplitude, which cross-cut the travel time distance trajectory (plotted in black) of the ray group under consideration.

c To effectively measure the accord between the recorded data with the proposed structure and wavespeed of the layer under consideration, the statistical coherence functional is Scalculated along the travel time distance curve of the ray group under consideration.

In this implementation of the method, the semblance functional S for each ray group is calculated along the trajectory in the data defined by the travel time distance curve. The functional is calculated over the extended range of offsets provided by wide aperture seismic data, as a measure of the multichannel coherence along this travel-time distance trajectory. For the observed seismic signal 'ij where i is the trace number of the observed data traces (totalling m) within the range of offsets bracketed by the ray group's travel time-distance curve, and j is the sample corresponding to the travel time at this offset derived from the piecewise continuous segments of each ray group's travel time-distance trajectory, and aij is the ray amplitude calculated at this offset and travel time S m(Y Y aij Tlij 2 C (aij tij) 2 In this implementation of the method, an initial model of each layer is constructed (the a priori model), using available information on the structure and seismic parameters. The parameters defining the layer in the model are iteratively improved using the Jacobian matrix. This Jacobian matrix is constructed by calculating this semblance functional for a range of different wavespeed parameters associated with each layer, and for a range of different geometries of the interfaces circumscribing the layer; differentially perturbing the a priori model and re-calculating the semblance functional. The Jacobian matrix is then used to maximize this semblance functional for the ray groups that reflect or refract within that layer, iteratively improving the a priori model.

This procedure is undertaken, in turn, for each common seismic gather, and each layer, working from shallowest to deepest layers, until an optimal model has been constructed to the desired maximum depth of interest.

Figure 11 illustrates travel time distance trajectories (plotted in black) for a range of W AS

J

primary compressional reflections (primary PP reflections) for interfaces in the Marmousi-2 model, generated for a seismic source at X=4475 m, and overplotted on the

U

O synthetic seismogram produced by finite difference computer code by Martin( 2004, MSc thesis, University of Houston).

Cl The resulting model provides a realistic representation of the subsurface structure of layers and seismic parameters, either or both compressional and shear wavespeed, f within each layer. The technique utilizes recorded wide aperture seismic data of the N compressional or shear wavefield. The technique avoids the need to gather and stack the C'q seismic data, thereby precluding the need to make assumptions inherent in such a c gathering and stacking operation. By so doing, the technique avoids problems entailed with the stacking of seismic gathers, and can effectively utilize the data recorded by seismic receivers at long offsets from the seismic source.

Cl The disclosed technique constructs the model directly as a depth model, and consequently, there is no requirement to subsequently migrate the derived model. The model provides accurate information on the depth of target layers of interest.

The disclosed technique provides an efficient methodology to build the subsurface model. The technique is capable of generating models of sufficient resolution for application in the search for hydrocarbons reserves in areas of structurally complex geology.

W A S

J

geIZ:::

Claims

2. A computer-implemented method as claimed in claim 1, in which the seismic model is either two or three dimensional. The model comprises layers, which may represent either sedimentary strata or non-sedimentary geological units. Each layer is bounded by interfaces defining the area (2 dimensional) or volume (3-dimensional) of that layer, and within which area or volume the seismic parameters of the compressional and shear wavespeed and the density of the layer are fully specified.
3. A computer-implemented method as claimed in claims 1 and 2, in which the step of calculating the raypath comprises calculating seismic travel path and travel time for each alternative seismic ray by solving the differential equations defining the ray path and W AS J P N' travel time in each layer of the model through which that ray transits. o 4. A computer-implemented method as claimed in claims 1 to 3, in which the plurality a of alternative ray paths, comprising either compressional or shear rays in each transited layer, are grouped and identified by those interfaces at which the ray reflects, or through Swhich the ray refracts, and by the nature of the ray in each transited layer. The rays are grouped accordingly, and identified as a member of a series of ray groups.
5. A computer-implemented method as claimed in claims 1 to 4, in which each selected ray group associated with the layer under consideration, having a common set of c interfaces at which the member rays have reflected, and a common set of layers within which the member rays have refracted, defines the travel time distance trajectory along kn which a statistical coherence functional of the observed wavefield is calculated for that ray group.
6. A computer-implemented method as claimed in claims 1 to 5, in which the parameters defining the layer under consideration in the model are differentially altered to evaluate the Jacobian matrix of the statistical coherence functional. Both the geometry parameters of the layer and the compressional and shear wavespeeds of the layer are perturbed from a priori values to calculate the partial derivative of this statistical coherence functional with respect to each parameter associated with each layer.
7. A computer-implemented method as claimed in claims 1 to 6, in which the layers are progressively constructed, by adjusting the layer parameters to maximize the statistical coherence functional averaged over each seismic gather for the ray group and layer under consideration, using the Jacobian matrix to establish, in a progressive manner, the optimum geometry and seismic parameters of each layer. W AS