WO2024016047A1

WO2024016047A1 - Method for the segmentation of seismic data of a subsurface location

Info

Publication number: WO2024016047A1
Application number: PCT/AU2023/050599
Authority: WO
Inventors: Jeremy Barrett Smith; Andrew Meng Leong Pethick
Original assignee: HiSeis Pty Ltd
Priority date: 2022-07-20
Filing date: 2023-06-29
Publication date: 2024-01-25

Abstract

The present invention relates to a method for the segmentation of seismic data of a subsurface location, the method comprising: providing seismic data of a subsurface location; processing the seismic data to generate a set of seismic textural attributes; processing the set of seismic textural attributes to generate vectorized attribute data; and performing a segmentation analysis on the vectorized attribute data to produce segmented seismic data.

Description

Method for the Segmentation of Seismic Data of a Subsurface Location

TECHNICAL FIELD

[0001 ] In accordance with the present invention, there is provided a method for the segmentation of seismic data of a subsurface location. More specifically, forms of the present invention provide for a method for the seismic data segmentation using the textural attributes of the subsurface location.

BACKGROUND ART

[0002] The following discussion of the background art is intended to facilitate an understanding of the present invention only. The discussion is not an acknowledgement or admission that any of the material referred to is or was part of the common general knowledge as at the priority date of the application.

[0003] Geophysical data is used to provide information on the physical properties of subsurface locations. Geophysical data is primarily used in mineral exploration to identify mineral deposits. The collected geophysical data is processed to characterize the physical properties of the subsurface location, which are then interpreted to infer geological information. Different types of geophysical data can be generated to characterize different physical properties of the subsurface location. Geophysical data types include electromagnetic data, magnetic data, seismic data, radiometric data and gravimetric data. Regardless of the type, the generated geophysical data must be correctly interpreted to properly characterize the physical properties. To assist with interpretation, computer algorithms are used to process geophysical data to infer distributions of subsurface physical properties. This process is often referred to as ‘inversion’. Inversion is used to fit the recorded geophysical data to a physical model. The physical model is not a geological interpretation or an estimate of lithology, but rather a representation or estimation of the distribution of subsurface physical properties. This physical model is then used to assist geological interpretation. The primary challenge in geophysical data inversion is that an infinite number of physical models can often explain a particular set of measurements. This is referred to in the art as non-uniqueness. New or improved data processing techniques are required to assist with geological interpretation. [0004] The most prevalent form of geophysical data used in the characterization of subsurface locations is reflective seismic data (hereafter “seismic data”). The collection of seismic data requires the generation of seismic (acoustic) waves and the measurement of the reflection of the seismic waves. These reflections are recorded as a function of time in many locations for a survey. The amplitude and timing of the reflected waves are a function of the subsurface acoustic impedance which is the product of a material’s compressional velocity and density, which is often related to the physical properties of the subsurface location, such as rock type, hydrocarbon content, and saturation. The seismic data is processed to fit the data to a physical model, which is used in assisting geological interpretation.

[0005] To improve the geological interpretation of seismic data, a number of seismic attributes can be extracted from the seismic data. Traditional seismic attributes are generally based on measurements of time, amplitude, frequency, and/or attenuation. Seismic attributes assist to identify or highlight geophysical features, relationships, and patterns that otherwise might not be noticed.

[0006] Throughout this specification, unless the context requires otherwise, the word "comprise" or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers.

SUMMARY OF INVENTION

[0007] In accordance with a first aspect of the present invention, there is provided a method for the segmentation of seismic data of a subsurface location, the method comprising: providing seismic data of a subsurface location; processing the seismic data to generate a set of seismic textural attributes; processing the set of seismic textural attributes to generate vectorized attribute data; and performing a segmentation analysis on the vectorized attribute data to produce segmented seismic data.

[0008] The inventors have found that the method of the present invention can be used to classify geophysical data into zones of materials with similar textural features. The segmented seismic data may then be used to construct a structural model of the subsurface location. The inventors have found that by generating a set of seismic textural attributes and performing a segmentation analysis, subjectivity within the interpretation of the seismic data is reduced. This in turn reduces the time taken to construct a structural model. The segmented seismic data may also be used in the identification and delineation of large intrusive bodies, identification of fluid pathways and detailed mapping of near-surface conditions. The segmented seismic data may also be used to aid in the interpretation of other geophysical data.

[0009] Throughout this specification, unless the context requires otherwise, the term "structural model" or variations, will be understood to refer to a representation of the geological structures of the subsurface location. Geological structures include boundaries between rock units, faults, shear zone location and the dips of geological bedding or structures. Structural models are useful to understand the deformation history of the target area and can be used to map fluid pathways, geohazards and depth maps to geological boundaries.

[0010] In one form of the present invention, the method comprises the generation of the seismic data. Alternatively, historical seismic data may be processed.

[001 1 ] In one form of the present invention, the seismic data is two-dimensional (2D) or three-dimensional (3D) seismic data. Preferably, the seismic data is stored within the SEG-Y format.

[0012] In one form of the present invention, the method comprises the step of: pre-processing the seismic data prior to the step of processing the seismic data to generate a set of seismic textural attributes. [0013] In one form of the present invention, pre-processing the seismic data comprises filtering the seismic data. Preferably, the seismic data is filtered to clip the data values to within a specified amplitude range. Additionally or alternatively, pre-processing the seismic data comprises the conversion of the seismic data into discrete intensity levels.

[0014] In one form of the present invention, the set of seismic textural attributes comprise one or more of: Grey Level Co-Occurrence Matrix (GLCM) feature attributes, Grey Level Run Length Matrix (GLRLM) feature attributes, Grey Level Size Zone Matrix (GLSZM) feature attributes, Grey Level Dependence Matrix (GLDM) feature attributes, and Neighbouring Gray Tone Difference Matrix (NGTDM) feature attributes.

[0015] In one form of the present invention, one or more secondary attributes are incorporated into the vectorized attribute data.

[0016] In one form of the present invention, the secondary attributes comprise non- textural seismic attributes.

[0017] In one form of the present invention, the secondary attributes comprise petrophysical attributes.

[0018] In one form of the present invention, processing the set of seismic textural attributes comprises the removal of unsuitable seismic textural attributes. Preferably, the unsuitable seismic textural attributes include those that contain errors, those not deemed trustworthy and those which do not recover sufficient textural information.

[0019] In one form of the present invention, processing of the set of seismic textural attributes includes assigning a weighting to each seismic textural attribute.

[0020] In one form of the present invention, the method comprises the step of: normalising the vectorized attribute data, prior to the segmentation analysis.

[0021 ] In one form of the present invention, a dimensionality reduction is applied to the vectorized attribute data prior to the segmentation analysis. [0022] In one form of the present invention, the segmentation analysis comprises the processing of the vectorized attribute data using one or more segmentation algorithms.

[0023] In one form of the present invention, machine learning is used in the segmentation analysis. Preferably, unsupervised learning is used in the segmentation analysis.

[0024] In one form of the present invention, the method further comprises the step of: constructing a structural model of the subsurface location using the segmented geophysical seismic data.

[0025] In one form of the present invention, the subsurface location comprises solid rock. Preferably, the subsurface location comprised hard rock.

[0026] In accordance with a second aspect of the present invention, there is provided a system for the segmentation of seismic data of a subsurface location, the system comprising: a controller; storage storing electronic program instructions for controlling the controller; and an input means; wherein the controller is operable, under control of the electronic program instructions, to: receive input via the input means, the input comprising seismic data; process the input to generate an output set of seismic textural attributes; process the output set of seismic textural attributes to generate vectorized attribute data; process the vectorized attribute data using a segmentation analysis to generate an output segmented seismic data; and communicate the output segmented seismic data. [0027] In one form of the present invention, the system comprises a display for displaying a user interface, and the controller is operable, under control of the electronic program instructions, to communicate the output by displaying the output via the display.

[0028] In one form of the present invention, the system comprises a display for displaying a user interface, the controller is operable, under control of the electronic program instructions, to communicate the output set of seismic textural attributes by displaying the output via the display.

[0029] In one form of the present invention, the controller is operable, under control of the electronic program instructions, to allow a user to assign weightings to the set of seismic textural attributes.

[0030] In one form of the present invention, the controller is further operable, under control of the electronic program instructions, to perform a pre-processing operation on the input seismic data. Preferably, the pre-processing operation comprises a filtering operation. Preferably, the pre-processing operation comprises a conversion operation to output seismic data at discrete intensity levels.

[0031 ] In one form of the present invention, the controller is further operable, under control of the electronic program instructions, to receive secondary attribute as input data and combine the input secondary attributes with the vectorized attribute data.

[0032] In one form of the present invention, the controller is further operable, under control of the electronic program instructions, to perform removal operation on the set of seismic textural attributes to remove unsuitable textural attributes.

[0033] In one form of the present invention, the system comprises a display for displaying a user interface, the controller is operable, under control of the electronic program instructions, to communicate the output set of seismic textural attributes by displaying the output via the display.

[0034] In one form of the present invention, the controller is operable, under control of the electronic program instructions, to allow a user to assign weightings to the set of seismic textural attributes. [0035] In one form of the present invention, the controller is further operable, under control of the electronic program instructions, to perform a normalising operation on the vectorized attribute data.

[0036] In one form of the present invention, the controller is further operable, under control of the electronic program instructions, to perform a dimensionality reduction operation on the vectorized attribute data.

[0037] In one form of the present invention, the controller, under control of the electronic program instructions, processes the vectorized attribute data using one or more segmentation algorithms.

[0038] In one form of the present invention, the controller, under control of the electronic program instructions, is operable to leverage machine learning in the segmentation analysis.

[0039] The system may be implemented in a device. The device may be a mobile communication device, in which case it may comprise a smartphone, notebook/tablet/desktop computer, or portable media device, having the electronic program instructions, which may comprise software, installed thereon. The software may be provided as a software application downloadable to the device, and/or running on servers and/or the cloud as a service.

[0040] Preferably, one or more operations performed by the system occur automatically, without requiring human intervention.

[0041 ] In accordance with a third aspect of the present invention, there is provided a computer-readable storage medium on which is stored instructions that, when executed by a computing means, causes the computing means to perform the method according to the first aspect of the present invention as hereinbefore described.

[0042] In accordance with a fourth aspect of the present invention, there is provided a computing means programmed to carry out the method according to the first aspect of the present invention as hereinbefore described. [0043] In accordance with a fifth aspect of the present invention, there is provided a data signal including at least one instruction being capable of being received and interpreted by a computing system, wherein the instruction implements the method according to the first aspect of the present invention as hereinbefore described.

[0044] In accordance with a sixth aspect of the present invention, there is provided a device for the segmentation of seismic data of a subsurface location comprising a system according to the second aspect of the present invention as hereinbefore described.

[0045] One embodiment provides a computer program product for performing the method according to the first aspect of the present invention as hereinbefore described.

[0046] One embodiment provides a non-transitive carrier medium for carrying computer executable code that, when executed on a processor, causes the processor to perform the method according to the first aspect of the present invention as hereinbefore described.

[0047] One embodiment provides a system configured for performing the method according to the first aspect of the present invention as hereinbefore described.

BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Further features of the present invention are more fully described in the following description of two non-limiting embodiments thereof. This description is included solely for the purposes of exemplifying the present invention. It should not be understood as a restriction on the broad summary, disclosure or description of the invention as set out above. The description will be made with reference to the accompanying drawings in which:

Figure 1 is a schematical representation of the method for the segmentation of seismic data of a subsurface location in accordance with an embodiment of the present invention;

Figure 2 shown the results of various stages of the process of Figure 1 ; and

Figure 3 shows the impact that individualized weighting the seismic textural attributes has on the segmentation outcome. DESCRIPTION OF EMBODIMENTS

[0049] The method of the present invention relates generally to a method for the segmentation of seismic data of a subsurface location. The method analyses the seismic data to extract a set of seismic textural attributes from the seismic data. The set of seismic textural attributes is used to segment the data into zones having similar textural properties. The segmented data may then used to construct a structural model of the subsurface location which is segmented according to these zones. Alternatively, the segmented data may be used to guide interpretation of geophysical data or similar applications.

[0050] In a preferred embodiment of the present invention, the seismic data is seismic reflection data or a derivative thereof. Seismic reflection data is obtained through the generation of seismic waves through the subsurface location and measuring reflections of the seismic waves as a function of time. The seismic waves are reflected off interfaces of acoustic impedance contrasts. Seismic reflection data may be two-dimensional (2D) or three-dimensional (3D). The output SEG-Y binary header and binary trace header information are used to form geometry to either create a rectilinear grid (2D) or a rotated 3D rectilinear volume positioned in real-world coordinates. As would be appreciated by a person skilled in the art, seismic reflection data is typically in the form of the SEG-Y data exchange format as outlined by the SEG Technical Standards Committee (see https://seq.ora). It is envisaged by the inventors that other formats of seismic data may similarly be used in the method of the present invention. Similarly, it is envisaged that other formats of seismic data may be converted into SEG-Y format.

[0051 ] Figure 1 shows a flowsheet of a method for the segmentation of seismic data of a subsurface location in accordance with an embodiment of the present invention. In one embodiment, the subsurface location is solid rock. Preferably, the subsurface location is hard rock.

Obtaining Seismic Data

[0052] In one embodiment of the present invention, the method includes the step of generating seismic data. It is assumed by the inventors that the present invention will be most useful for use in the geological mapping of potential mineral deposits for geological exploration. To conduct such mapping, seismic data of the potential mineral deposit will need to be generated. The method of generating the seismic data will depend on the type of seismic data and those skilled in the art would be aware of the techniques for generating such data. In forms of the invention where the seismic data is seismic data, the step of generating seismic data comprises propagating seismic waves through the subsurface location and recording time and amplitude of the reflected seismic waves. It is further envisaged that historical seismic data may be processed using the method of the present invention. The raw seismic data is preferably subjected to routine processing methodologies known in the art to improve resolution, including stacking, time/depth migration and deconvolution. The processed seismic data is typically referred to as a stacked seismic section or stacked seismic volume.

[0053] The obtained seismic data 101 is processed in accordance with the method of the present invention.

Pre-processing the Seismic Data

[0054] In one embodiment, the method comprises the step of pre-processing the seismic data 102. The pre-processing step 102 aims to condition or normalise the input seismic data prior to further processing.

[0055] In one embodiment of the present invention, the pre-processing step 102 comprises filtering the seismic data. Preferably, the seismic data is filtered to clip the data values to within a specified amplitude range. Filtering of the seismic data removes outlier data below a minimum amplitude value and above a maximum amplitude value. The inventors have found that removing the highest negative and positive amplitude events will increase the dynamic range of the discretized seismic data, thereby assisting to represent low amplitude events in the seismic data. In a preferred embodiment, the clip for seismic amplitude data is between 0.1 and 5%. Those skilled in the art would appreciate there are multiple algorithms that can be applied to clip the amplitude of seismic data. One example is the alpha-trim clip. Preferably, the alpha-trim clip for seismic amplitude data is between 0.1 and 5%.

[0056] In one embodiment of the present invention, the pre-processing step comprises the conversion of the seismic data into discrete intensity levels. Preferably, the pre- processing step comprises the conversion of the seismic data into discrete gray level intensities. Gray level intensities are discrete integer values representing scalars. The number of gray levels should be carefully selected to ensure that both low-amplitude and high-amplitude events are represented without significant saturation while minimizing the number of gray levels for computational efficiency. More preferably, conversion of the seismic data into discrete intensity level intensities the seismic data comprises normalising the floating I double precision data into discrete integer values between 0 and N_g where, N_g is the number of gray levels. Each value, v, is normalised between the range of V_min to V_max, where GI is the resulting percentage. The normalisation method is shown in equation 1 . The “inf function rounds the value to the nearest whole number and casts the floating point to an integer primitive. This process is results in the generation of levelled seismic data.

GI = intf (1 )

[0057] In a preferred embodiment, the step of filtering the seismic data is conducted prior to the conversion of the seismic data into discrete intensity level intensities. Filtering removes outliers and reduces the final range of range of V_min to V_max. The inventors have found filtering the data to remove outliers allows the conversion of discrete gray levels to be performed without saturating the higher or lower seismic amplitude reflection events. This filtering improves the resolution of the final discretized gray levels and in-turn both the resulting textures and clustering results.

Seismic Textural Attribute Generation

[0058] The levelled seismic data resulting from the pre-processing step is processed to generate a set of seismic textural attributes 103. In this step, the seismic data is analysed to extract a set of attributes based on different features of the seismic data. Preferably, a set of feature extraction algorithms are applied to the seismic data to generate a set of seismic textural attributes. Each seismic textural attribute is a measurable property of the seismic data which may be used to highlight or identify geological or geophysical features. Whilst individual seismic textural attributes may not provide sufficient information in isolation, the combination of a set of seismic textural attributes may be used to segment the seismic data into distinct regions that are compatible with a prescribed pattern. Those skilled in the art would recognize that existing seismic data analysis techniques incorporate the use of seismic attributes to segment or partition the data into geobodies or regions defined by similar seismic attributes. However, traditional seismic attributes do not extract textural attributes and are instead generally based on measurements of time, amplitude, frequency, and/or attenuation.

[0059] The method of the present invention requires the extraction of various attributes based on the textural properties of the subsurface location. The textural properties of a rock are dependent on the mineral species, grain size, shape, and orientation. Analysis of the textural properties of the seismic data has been found to be useful in assisting in the identification and characterization of the subsurface location being examined. In the field of seismic data analysis, seismic texture is a quantitative measure of the reflection amplitude, continuity, and internal configuration of reflectors. Seismic reflection imaging detects the contrast in subsurface acoustic impedance, being the contrast in the product of bulk density and acoustic velocity. Without wishing to be limited by theory, the present inventors understand that contrast in acoustic impedance will be uniform where contrasts in geological/lithological boundaries are similar. Rock mass volumes would therefore yield similar seismic texture.

[0060] Analysis of the seismic data can extract a number of different seismic textural attribute that are quantifiable through various statistical techniques. A set of standardized statistical techniques for generating a set of seismic textural attributes from medical imaging data is defined by the Imaging Biomarker Standardization Initiative (IBSI) as detailed in Zwanenburg, A., Leger, S., Vallieres, M., and Lock, S. (2016). “Image biomarker standardisation initiative - feature definitions", the contents of which are include by reference herein in their entirety. This group has standardized the extraction of image biomarkers from medical imaging for purposes within the field of quantitative image analysis. The inventors of the present invention have modified the techniques described in IBSI for use with seismic data, as follows: i. The ISBI techniques specify that texture feature sets require interpolation to isotropic voxel spacing to be rotationally invariant. This allows comparison between image data from different samples cohorts or batches. Given the large ratio between horizontal and vertical sampling within geophysical seismic data, isotropic voxel spacing cannot be implemented without suffering a large computational overhead or a sacrifice in resolution. To overcome this, all data for textural comparison will be projected on the same specified seismic mesh using an appropriate interpolant function known in the art. Suitable interpolant functions known in the art include nearest neighbour, bilinear/trilinear splines and 2D/3D kriging interpolator. ii. The ISBI techniques implement region of interest (ROI) morphological and intensity mask attributes. The inventors have found that the high level or variability in the range, size, structure properties and morphological properties of subsurface locations means that seismic data is too complex for computing attributes within a ROI mesh. Instead of computing attributes within a ROI mesh, the attributes are computed for each sample of the SEGY volume within a window.

[0061 ] In one embodiment of the present invention, the set of seismic textural attributes comprise one or more of: Grey Level Co-Occurrence Matrix (GLCM) feature attributes, Grey Level Run Length Matrix (GLRLM) feature attributes, Grey Level Size Zone Matrix (GLSZM) feature attributes, Grey Level Dependence Matrix (GLDM) feature attributes, and Neighbouring Gray Tone Difference Matrix (NGTDM) feature attributes. In one embodiment of the present invention, the set of seismic textural attributes comprise two or more of: GLCM feature attributes, GLRLM feature attributes, GLSZM feature attributes, GLDM feature attributes, and NGTDM feature attributes. In one embodiment of the present invention, the set of seismic textural attributes comprise three or more of: GLCM feature attributes, GLRLM feature attributes, GLSZM feature attributes, GLDM feature attributes, and NGTDM feature attributes. In one embodiment of the present invention, the set of seismic textural attributes comprise four or more of: GLCM feature attributes, GLRLM feature attributes, GLSZM feature attributes, GLDM feature attributes, and NGTDM feature attributes. In one embodiment of the present invention, the set of seismic textural attributes comprise each of GLCM feature attributes, GLRLM feature attributes, GLSZM feature attributes, GLDM feature attributes, and NGTDM feature attributes.

[0062] Grey Level Co-occurrence Matrix (GLCM) based feature attributes are a defining set of features calculated from the probability matrix which is the GLCM. This matrix expresses the relationship of each pixel/voxel computed grey levels (a specified number of discrete intensities) to their neighbouring pixels/voxels. The IBSI details twenty-four (24) GLCM feature attributes that may be implemented. In embodiments where the set of seismic textural attributes comprises GLCM feature attributes, the set of seismic textural attributes may contain one or more of the 24 GLCM feature attributes. To assist in the processing of seismic data, the inventors have found that the standard GLCM matrix calculation may be modified to allow for directional weighting. Where each GLCM matrix computation for each direction has its own independent weighting (W). This makes the GLCM matrix a directionally weighted probability matrix. The weightings can be assigned automatically for each voxel location using dip, azimuth and dip and azimuth confidence attribute volumes. Those skilled in the art would appreciate these attributes may be computed using commercially available seismic interpretation packages, including, for example, DUG insight, Opendtect or Petrel. The dip and azimuth and associated confidence attributes provide a directional vector of planar dip perpendicular to each reflector. The weighting matrix can be designed according to this, by biassing the direction of dip parallel to the plane of the dip. Another approach is to emphasise features in specific directions, whiles the GLCM matrix maintaining a contribution, but lesser, from other orientations. For example, if a user wishes to bias GLCM attribute generation twice as much in the vertical orientation to highlight continuous vertical reflectors, then a weighting in the vertical directions will be 1 , while the horizontal weighting will be 0.5. The GLCM matrix, M, is a N x N matrix, where N is the number of grey levels. The standard GLCM matrix M, is comprised of integer values where each adjacent cell contributes 1 to the corresponding cell within the GLCM matrix. When independent weighting is implemented, each adjacent cell contributes the weighting W to the directional GLCM matrix. In this example the directional GLCM matrix, M_m+=^, has been given a weighting of 0.5. The final GLCM matrix will merge all of the results together prior to feature attribute generation.

Standard

Weighted

In this example the directional GLCM M_m+=^ has a weighting of 0.5. j j

1 2 2 3 0 1.5 6 6" 0 0 0 2

/(i,j) = l 2 3 3 M_m+=^ . 0 0.5 1.5 0.5 M_m-= _ = . 3 1 0 1

I I

4 2 4 1 0 1 0.5 0 0 3 1 0

4 1 2 3 1 0.5 0 0 0 1 0 0

[0063] Grey Level Run Length Matrix (GLRLM) feature attributes assesses the distribution of grey levels along specified directions of neighbouring voxels. A run length is defined as a length of consecutive pixels with the same discretised intensity in the specified direction. GLRM matrices can be computed independently for each direction or merged according to the aggregation methodology (IAZD). The IBSI details sixteen (16) GLRLM feature attributes may be implemented. In embodiments where the set of seismic textural attributes comprises GLRLM feature attributes, the set of seismic textural attributes may contain one or more of the 16 GLRLM feature attributes. No substantial modification to the methodology described by the IBSI is required to process seismic data.

[0064] Grey Level Size Zone Matrix (GLSZM) feature attributes assesses the texture related to the size of the interconnectivity of voxels. The matrix measures the number of zones of linked voxels sharing the same grey level intensity. This algorithm has implemented recursive test for interconnectivity along all 26 neighbouring voxels for 3D data and 8 voxels for 2D data. The IBSI details sixteen (16) GLSZM feature attributes may be implemented. In embodiments where the set of seismic textural attributes comprises GLSZM feature attributes, the set of seismic textural attributes may contain one or more of the 16 GLSZM feature attributes. No substantial modification to the methodology described by the IBSI is required to process seismic data.

[0065] Grey Level Dependence Matrix (GLDM) feature attributes assesses how connected voxels within a specified Chebyshev distance are dependent on the centre voxel. A cell with a gray level intensity, j, is considered dependent on the centre voxel if the gray level intensity, i, is less than an alpha value, a. The full derivation of the matrix is shown in (2). That is, it is considered dependent if |i - j| < a. The IBSI details fourteen (14) GLDM feature attributes may be implemented. In embodiments where the set of seismic textural attributes comprises GLDM feature attributes, the set of seismic textural attributes may contain one or more of the 14 GLDM feature attributes. No substantial modification to the methodology described by the IBSI is required to process seismic data.

[0066] Neighbouring Gray Tone Difference Matrix (NGTDM) feature attributes are a statistical different measure between a gray level intensity and its neighbours within a specified Chebyshev distance. The IBSI details five (5) NGTDM feature attributes may be implemented. In embodiments where the set of seismic textural attributes comprise NGTDM feature attributes, the set of seismic textural attributes may contain one or more of the 5 NGTDM feature attributes. No substantial modification to the methodology described by the IBSI is required to process seismic data.

[0067] A full list of the possible seismic textural attributes that may be incorporated in the present invention are provided in Table 1 .

Table 1: Seismic textural attributes

Secondary Attributes

[0068] The inventors of the present invention have found that the segmentation of the seismic data may be improved with the inclusion of one or more secondary attributes. In this embodiment, the secondary attributes are combined with the set of seismic attributes prior to segmentation.

[0069] In one embodiment of the present invention, the secondary attributes comprise one or more standard seismic attribute sets 107. The extraction and processing of standard non-textural seismic attributes is well known in the art. A detailed discussion of standard seismic attributes is provided in Chopra, S. and Marfurt, K., 2007. Seismic Attributes for Prospect Identification and Reservoir Characterization. [0070] In one embodiment of the present invention, the secondary attributes comprise one or more petrophysical attribute sets 106. As would be appreciated by those skilled in the art, petrophysical data is the physical property of the subsurface material, which can include electrical conductivity, bulk density and acoustic velocity. Distributions of these attributes can be estimated using a variety of geophysical methods, including (6002) bulk density imaged by gravity and gravity gradiometry surveying (6004) distributions of subsurface electrical conductivity imaged by controlled source electromagnetics, marine controlled source electromagnetics, airborne, ground and borehole electromagnetic, passive source electromagnetics including audio magnetotellurics and magnetotellurics or direct-current resistivity methods, (6006) changes in subsurface chargeability imaged via induced polarization methods and (6001 ) distributions velocity detected through passive seismic and seismic tomography methods. Further details are provided in Table 2:

Table 2: Petrophysical attributes

Attribute Selection

[0071 ] In one embodiment of the present invention, the generated set of seismic textural attributes are subjected to a quality control step (105). The quality control step is used to identify seismic textural attributes which are unsuitable for later segmentation analysis either because (i) they contain errors (ii) they are not deemed trustworthy or (iii) they do not recover sufficient textural information. The identified seismic textural attributes are removed from the set of seismic textural attributes. [0072] In embodiments, where secondary attributes are included in the segmentation analysis, the set of secondary attributes may also be subjected to a quality control step. This may not be required in embodiments where the set of secondary attributes has already been processed.

Weighting

[0073] In one embodiment, each attribute (a_n) is assigned a numerical weighting (Wn). The assigned numerical weighting signifies the importance I influence of the attribute to the segmentation outcome. This weighting is used in the distance calculation function for each segmentation algorithm.

[0074] In one embodiment, each attribute assigned a weighting value between 0 and 1. Where 0 is assigning the attribute to have no influence on the segmentation outcome and 1 is assigning the attribute to have full influence. By default, all attributes are assigned a weighting of 1. These weighting parameters can be tuned to improve segmentation outcomes or to favour segmentation based on particular geological characteristics. In one embodiment the weighting of each attribute is defined by a user. Preferably, an assessment of each attribute is made based on the geological objective of the segmentation analysis and the geological characteristics highlighted by each attribute, being either a seismic textural attribute or a secondary attribute. For example, a geological objective designed to segment intrusive bodies will place weightings favouring attributes that differentiate this nature of geological characteristics over attributes that provide poor textural differentiation on this geological characteristic. Additionally or alternatively, the weighting of each attribute is defined by an algorithm. Preferably, an error minimization algorithm is applied to optimise these weighting parameters. It is envisaged that an error minimisation algorithm could be used to minimise the error between the segmentation outcomes and other geophysical data of the subsurface location, such as lithological or petrophysical well logs. Similarly, the choice of comparison data will depend on the particular geological outcome sought.

[0075] Those skilled in the art would appreciate how to apply weights to a distance function. For example, the calculation of the L1 and L2 Euclidean norm distance from each centroid for each single point for both unweighted, |x|, and weighted, |x_w| are shown for each attribute axis (n).

N ard L 1 Norm) is replaced by Weighted LI Norm) ndard L 2 Norm) is replaced by (Weighted L2 Norm)

Generating vectorized attribute data.

[0076] The set of seismic textural attributes, together with any secondary attributes, are subjected to an attribute post processing step (108) to generate vectorized attribute data. In one embodiment, the attribute post processing step (108) comprises mapping the seismic textural attributes onto a common domain. This is performed by (1 ) selecting the output spatial geometry from an input seismic dataset. This is used as the common geometry to which all other attributes will map to, (2) either a nearest neighbour algorithm with a max search distance applied, or an inverse distance weighted interpolation. The mapping of the seismic textural attributes to a common domain has been found to assist with the incorporation of a secondary attributes into the vectorized attribute data and the segmentation analysis. Once all of the attributes are in the same domain, the resulting attributes are processed to convert the data into a vectorized form (A), which details each attribute (a_n) for each point (m) as shown below in Table 3. Where each attribute (n) can be either a textural attribute or a secondary attribute such as a traditional seismic attribute or a petrophysical attribute. As would be appreciated by a person skilled in the art, vectorization generally comprises to two main steps. Firstly, for each attribute, all elements in the two-dimensional grid or three-dimensional volume is collapsed into a one- dimensional array (e.g., column ai in Table 3). This is also known multi-dimensional matrix flattening. Secondly, these individual column arrays are then stacked via a column stack to form the final attribute table (Table 3). Each row of the new matrix, A, is a vector comprising of n dimensions.

Table 3: Attribute Table after step 105.

Data Normalisation

[0077] In one embodiment, the method further comprises the step of normalising the vectorized attribute data 109. Data normalisation between different attributes prior to any dimensionality reduction or segmentation analysis has been found to assist in preventing the magnitude of the amplitude from biassing the results towards the higher magnitude attributes. Normalising typically includes two steps: (1 ) trimming the data to remove extremes. This is performed by calculating specified centile ranges. These ranges are typically around the 0.1 ^th and 99.9^th centile ranges, but can be user specified. The minimum (vmin) and maximum (v_max) centile values are then used to (2) normalise each data point (v) between 0 and 1 .

where V is the normalised value between 0 and 1 .

[0078] A normalised n by m matrix, A, is created where there are n attributes and m voxels within the section / volume. A is the matrix shown in table 3 and A is normalised matrix, where each column (n) is normalised between 0 and 1 .

Dimensionality Reduction

[0079] In one embodiment, a dimensionality reduction step (1 10) is applied to at least part of the vectorized attribute data 109. This is an optional step that may be applied if the number of attribute input variables exceed what is reasonable for the segmentation algorithm. As would be appreciated by a person skilled in the art, dimensionality reduction is a group of techniques that reduces the number of variables (i.e., attributes). For example, using the above matrix A, the vector dimension, n, is reduced to a smaller number. The output dimensionality reduced attribute matrix will be referred to as A*. The aim of dimensionality reduction is to facilitate effective clustering of higher-dimensionality datasets. As discussed above, the inventors of the present invention have identified upwards of 75 different seismic textural attributes that may be extracted from the seismic data. If the number of dimensions is too high, the segmentation analysis may not yield sensible results as the distance function that is applied may not capture higher order patterns and relationships. If higher-dimensionality problems are encountered, these dimensionality reduction algorithms may be applied (1 10) on matrix A. The dimensionality reduction may be applied to the entire matrix A, or may be applied to a smaller subset of the attributes. Multiple dimensionality reduction can be applied to different subsets to reduce the overall number of dimensions. Suitable dimension reduction algorithms include (i) pairwise Controlled Manifold Approximation (Wang et al. Journal of Machine Learning Research 22 (2021 ) 1 -73), t-SNE (van der Maaten and Hinton, Journal of Machine Learning Research, 9 (2008) 2579-2605), UMAP (Mclnnes et al., Mclnnes et al., arXiv e-prints, art. arXiv:1802.03426 (2018), and TriMap (Amid and Warmuth, arXiv e-prints, art. arXiv:1910.00204 (2019)). An example of PaCMAP dimensionality reduction applied to seven inputs, including textural and petrophysical data (203-209), is shown in 210. 210 is shown for illustration. While only one example is provided, the PacMAP algorithm outputs a number of axis which can be from zero to less than the number of inputs (0 to n-1 ). These output axis are then used as the inputs in the Segmentation stage (step 11 1 ). It is envisaged that the segmentation analysis may be performed on the vectorized attribute data both with and without the dimensionality reduction having been applied. Furthermore, it is envisaged that vectorized attribute data having different dimensionality reduction algorithms applied may be incorporated into the segmentation analysis.

Segmentation Analysis

[0080] A segmentation analysis (1 1 1 ) is then performed on the vectorized attribute data. The segmentation analysis is used to classify multi-dimensional point-sets. Segmentation assigns each spatial voxel a cluster reference. This cluster reference is an integer value. Each voxel which has the same reference shares similarities in seismic textural attributes and the secondary attributes. The segmentation analysis incorporates one or more segmentation algorithms. Each algorithm classifies each multi-dimensional point-set differently and will yield different results depending on the statistical relationship between the different seismic textural attributes and secondary attributes. Many different segmentation algorithms are known to those skilled in the art, suitable segmentation algorithms include, K-means/K-Means-i-i- (Kanungo et al., 2002. IEEE transactions on pattern analysis and machine intelligence, 24(7), pp.881-892; Vassilvitskii et al., 2006, In Proceedings of the eighteenth annual ACM-SIAM symposium on Discrete algorithms (pp. 1027-1035); and Maitra et al, 2010. A systematic evaluation of different methods for initializing the k-means clustering algorithm), G-means (Hamerly et al., 2003. Learning the k in k-means, Advances in neural information processing systems, 16.), Density Clustering (DENCLUE) (Hinneburg et al., 2003. Knowledge and information systems, 5(4), pp.387-415.; and Hinneburg et al., 2007, International symposium on intelligent data analysis (pp. 70-80)), Clustering Large Applications based upon Randomized Search (CLARANS) (Ng et al., 2002. IEEE transactions on knowledge and data engineering, 14(5), pp.1003-1016.), Density-Based Spatial Clustering of Application with Noise (DB Scan) (Hinneburg et al., 1996. KDD Conference, 1996; and Sander et al., 1998. Data mining and knowledge discovery, 2(2), pp. 169-194.).

[0081 ] In one embodiment, segmentation algorithms (step 11 1 ) are applied on either the matrix A" or A*. The output will be a single array of length m that contains classified segments as an integer. During step 11 1 , these algorithms have been applied to a subset or the whole matrix A or A*. The whole dataset from A and A* are then predicted from the computed clustering centroid model assigning a cluster each data point m from each element, m.

[0082] In one embodiment, machine learning is used in the segmentation analysis. Preferably, unsupervised machine learning is used in the segmentation analysis. As would be appreciated by a person skilled in the art, unsupervised machine learning is a topic of machine learning which learns patterns from untagged or uncategorized data. This is conducted without the requirement of human intervention, resulting in an unbiased categorization or segmentation of data. Unsupervised learning methods on the vectorized attribute data to find similarities across multiple co-located geophysical datasets, including the seismic textural attributes and the secondary attributes.

[0083] A number of segmentation algorithms may be trialled during the segmentation analysis to produce multiple segmentation outcomes. These segmentation outcomes may be compared to determine best clustering model to use. However, comparison of multiple segmentation outcomes may be difficult due to poor interpretability between segmentation datasets. Each segmentation outcome consists of clusters that are indexed according to areas of similar texture and the indexed order of each cluster is randomly assigned. The indexed order of the clusters of each segmentation outcome is independent to other segmentation outcomes, making it difficult to make a direct comparison. In one form of the present invention, the indexed order of each cluster is arranged according to the mean value of a user selected attribute. Arranging the clusters is accomplished by first performing a segmentation analysis in which the indexed order of each cluster is randomly assigned. The mean value of the user selected attribute for each indexed cluster is then calculated. The indexed order of each cluster may then be rearrange according to the calculated mean value of the user selected attribute. The user selected attribute may be any co-located data attribute associated with the seismic data. Suitable user selected attributes may be seismic energy, instantaneous frequency, and RMS amplitude. While these are common seismic attributes, it is envisaged that any seismic or textrual attribute can be used for order purposes. For example, a user may have computed a segmentation outcome for three clusters, c1 , c2 and c3 whereby the indexed order of each cluster is randomly assigned. For each voxel/cell of the clustered outcome, there is a corresponding seismic energy value. The mean of all seismic energy values for clusters c1 , c2 and c3 are then computed. If the mean values were found to be 1 .5, 0.5 and 1 .0 respectively for each cluster, the output cluster indices would be ordered as, c2, c3 and c1 . New ordered cluster indices are then assigned in the final output. Where c2 is now mapped as index 1 , c3 is mapped as index 2 and c1 as index 3. By implementing a similar indexed order for each segmentation outcome, the interpretability between multiple segmentation outcomes may be improved.

Output

[0084] The segmented seismic data is then used to generate a structural model of the subsurface location. This requires the clusters array to be reshaped back into the target geometry within the common domain. The results can then be exported into the standard SEGY format (112). Examples of the KMeans clustering for A* and A are shown in 211 and 212 respectively.

[0085] It is envisaged that the segmented seismic data or the structural model may be useful in geological mapping or to assist in the interpretation of other geophysical data, for example:

(a) Estimating attribute-based hard-rock geological signatures. The signature of either geological/mineralization targets, hosts, mineral pathways or mineral sources defined in terms of seismic texture or petrophysical properties.

(b) Creation and exporting of geo-bodies. Using the classified data to create and export a three-dimensional or two-dimensional wireframe mesh of the contained region.

(c) Aid seismic interpretation through simplification of seismic rock mass properties. (d) To be used to define structures for use in constrained inversion regions which can be used for petrophysical analysis, it is envisaged that petrophysical attributes may be combined with the structural model and a further inversion could be completed. From this, a better, structurally constrained petrophysical inversion is then created. This process can be recursive, creating a structure, inverting and then rerunning the workflow to create an improved inversion results.

[0086] The present invention further relates to a system implementing in a device the method for the segmentation of seismic data of a subsurface location described above. The device comprises a plurality of components, subsystems and/or modules operably coupled via appropriate circuitry and connections to enable the device to perform the functions and operations herein described. The device comprises suitable components necessary to receive, store and execute appropriate computer instructions to perform the method for the segmentation of seismic data of a subsurface location in accordance with embodiments of the present invention.

[0087] The device comprises computing means which in this embodiment comprises a controller and storage for storing electronic program instructions for controlling the controller, and information and/or data; a display for displaying a user interface; and input means.

[0088] The controller comprises processing means in the form of a processor.

[0089] The storage comprises read only memory (ROM) and random access memory (RAM).

[0090] The device is capable of receiving instructions that may be held in the ROM or RAM and may be executed by the processor. The processor is operable to perform actions under control of electronic program instructions, as will be described in further detail below, including processing/executing instructions and managing the flow of data and information through the device.

[0091 ] In the embodiment, electronic program instructions for the device are provided via a single standalone software application (app) or module, and/or as a software development kit (SDK) to be included or executed from within other apps, and/or a service running on servers and/or the cloud(s). In the embodiment described, the app, and/or SDK and/or service can be downloaded from a website (or other suitable electronic device platform) or otherwise saved to or stored on storage of the device and/or executed via an Application Program Interface (API).

[0092] In preferred embodiments of the invention, the device is computing means such as a personal, notebook or tablet computer.

[0093] The device also includes an operating system which is capable of issuing commands and is arranged to interact with the app to cause the device to carry out actions including the respective steps, functions and/or procedures in accordance with the embodiment of the invention described herein. The operating system may be appropriate for the device.

[0094] The app, and other electronic instructions or programs for the computing components of the device, can be written in any suitable language, as are well known to persons skilled in the art. In embodiments of the invention, the electronic program instructions may be provided as stand-alone application(s), as a set or plurality of applications, via a network, or added as middleware, depending on the requirements of the implementation or embodiment.

[0095] In alternative embodiments of the invention, the software may comprise one or more modules, and may be implemented in hardware. In such a case, for example, the modules may be implemented with any one or a combination of the following technologies, which are each well known in the art: a discrete logic circuit(s) having logic gates for implementing logic functions upon data signals, an application specific integrated circuit (ASIC) having appropriate combinational logic gates, a programmable gate array(s) (PGA), a field programmable gate array (FPGA) and the like.

[0096] The computing means can be a system of any suitable type, including: a programmable logic controller (PLC); digital signal processor (DSP); microcontroller; personal, notebook or tablet computer, or dedicated servers or networked servers.

[0097] The processors can be any custom made or commercially available processor, a central processing unit (CPU), a data signal processor (DSP) or an auxiliary processor among several processors associated with the computing means. In embodiments of the invention, the processing means may be a semiconductor based microprocessor (in the form of a microchip) or a macroprocessor, for example.

[0098] In embodiments of the invention, the storage can include any one or combination of volatile memory elements (e.g. random access memory (RAM) such as dynamic random access memory (DRAM), static random access memory (SRAM)) and nonvolatile memory elements (e.g. read only memory (ROM), erasable programmable read only memory (EPROM), electronically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), tape, compact disc read only memory (CD-ROM), etc.). The storage may incorporate electronic, magnetic, optical and/or other types of storage media. Furthermore, the respective storage can have a distributed architecture, where various components are situated remote from one another, but can be accessed by the processing means. For example, the ROM may store various instructions, programs, software, or applications to be executed by the processing means to control the operation of the device and the RAM may temporarily store variables or results of the operations.

[0099] The use and operation of computers using software applications is well-known to persons skilled in the art and need not be described in any further detail herein except as is relevant to the present invention.

[00100] Furthermore, any suitable communication protocol can be used to facilitate connection and communication between any subsystems or components of the device, and other devices or systems, including wired and wireless, as are well known to persons skilled in the art and need not be described in any further detail herein except as is relevant to the present invention.

[00101 ] Where the words “store”, “hold” and “save” or similar words are used in the context of the present invention, they are to be understood as including reference to the retaining or holding of data or information both permanently and/or temporarily in the storage means, device or medium for later retrieval, and momentarily or instantaneously, for example as part of a processing operation being performed. [00102] Additionally, where the terms “system”, “device”, and “machine” are used in the context of the present invention, they are to be understood as including reference to any group of functionally related or interacting, interrelated, interdependent or associated components or elements that may be located in proximity to, separate from, integrated with, or discrete from, each other.

[00103] Furthermore, in embodiments of the invention, the word “determining” is understood to include receiving or accessing the relevant data or information.

[00104] The controller is operable, via execution of applications such as the app to collect and process user inputs, and receive data and/or information in different types or formats, pertinent to the segmentation of seismic data of a subsurface location.

[00105] The device comprises operably connected/coupled components facilitating performance and operations as described, including appropriate computer chips (integrated circuits), transceiver/receiver antennas, and software for the sensory technology being used.

[00106] A database or databank also resides (at least in part in embodiments) on the storage and is accessible by the controller under control of the app. The controller is arranged to interact with the database as appropriate to cause the device to carry out actions including the respective steps, functions and/or procedures in accordance with the embodiment of the invention described herein.

[00107] It will be understood that any of the database(s) described may reside on any suitable storage device, which may encompass solid state drives, hard disc drives, optical drives or magnetic tape drives. The database(s) described may reside on a single physical storage device or may be spread across multiple storage devices or modules.

[00108] The database is coupled to the controller and in data communication therewith in order to enable information and data to be read to and from the database as is well known to persons skilled in the art. Any suitable database structure can be used, and there may be one or more than one database. In embodiments of the invention, the database can be provided locally as a component of the device (such as in the storage) or remotely such as on a remote server, as can the electronic program instructions, and any other data or information to be gathered and/or presented.

Example 1

[00109] Figure 2 shows a series of seismic data that has been processed by the method of the present invention. Standard 2D seismic data of a subsurface location was obtained and used as the input data 201 . Input data 201 is then subjected to pre-processing routine in which outliers are removed and the intensity levels are discretised into discrete gray levels, to produce pre-processing output 202. The pre-processed output 202 was then processed to extract a set of seismic textural attributes based on the following statistical methods. The example workflow comprises of inputs generated from (i) traditional methods, "Average Energy” (203), (ii) the Seisomics attributes, GLCM F23 Sum Entropy computed in All Directions (204), GLSZM F05 Size Zone Non-Uniformity (205), GLRLM F07 Run Percentage computed in the vertical Orientation (206), NGTDM F02 Contrast All Directions (207), GLDM F05 Dependence Non-Uniformity (208) and (iii) Petrophsical Data, seismic tomographic velocity (209).

[001 10] The resulting set of seismic textural attributes (203 to 209) is processed using a dimension reduction algorithm PacMAP. In this example, this reduces the number of dimensions down from eight to two. One of the dimensions, Axis 1 , is shown in 210.

[001 1 1 ] The set of seismic textural attributes, both without dimensionality reduction and with dimensionality reduction was then subjected to a segmentation analysis using a K- means segmentation algorithm. The outputs of the K-means clustering for both data sets are shown in 21 1 and 212 respectively. The discrete colour of each region highlights where the groups of seismic reflectors share a common set of textural and petrophysical characteristics. From these outputs, geological understanding can be derived, as areas sharing similar rock-mass characteristics will share the same cluster index.

Example 2

Figure 3 shows the impact that the weighting of seismic textural attributes may have on the segmentation outcome. Seismic data (301 ) was processed and the following seismic textural attributes were derived: a1 GLCM UX (302), a2 GLDM F03 Gray Level Non Uniformity (304) and a3 NGTDM F04 Complexity. Segmentation outcome 305 prioritizes attribute 302, weighting attributes a1 , a2 and a3 as 1.0, 0.1 and 0.1 respectively. Segmentation outcome 306 prioritizes attribute 303, weighting attributes a1 , a2 and a3 with 0.1 , 1.0 and 0.1 respectively. Segmentation outcome 307 prioritizes attribute 304, weighting attributes a1 , a2 and a3 with 0.1 , 0.1 and 1.0 respectively. For comparison a segmentation outcome weighting all attributes equally is shown in 308. These examples show that by prioritizing certain seismic textural attributes, the data may be segmented to highlight particular geophysical features.

Claims

1 . A method for the segmentation of seismic data of a subsurface location, the method comprising: providing seismic data of a subsurface location; processing the seismic data to generate a set of seismic textural attributes; processing the set of seismic textural attributes to generate vectorized attribute data; and performing a segmentation analysis on the vectorized attribute data to produce segmented seismic data.

2. A method according to claim 1 , wherein the method comprises the step of: pre-processing the seismic data, prior to the step of processing the seismic data to generate a set of seismic textural attributes.

3. A method according to claim 2, wherein pre-processing the seismic data comprises filtering the seismic data.

4. A method according to claim 2 or 3, wherein pre-processing the seismic data comprises the conversion of the seismic data into discrete intensity levels.

5. A method according to any of the preceding claims, wherein the set of seismic textural attributes comprise one or more of: Grey Level Co-Occurrence Matrix (GLCM) feature attributes, Grey Level Run Length Matrix (GLRLM) feature attributes, Grey Level Size Zone Matrix (GLSZM) feature attributes, Grey Level Dependence Matrix (GLDM) feature attributes, and Neighbouring Gray Tone Difference Matrix (NGTDM) feature attributes.

6. A method according to any of the preceding claims, wherein one or more secondary attributes are incorporated into the vectorized attribute data.

7. A method according to any of the preceding claims, wherein processing the set of seismic textural attributes comprises the removal of unsuitable seismic textural attributes.

8. A method according to any of the preceding claims, wherein processing of the set of seismic textural attributes includes assigning a weighting to each seismic textural attributes.

9. A method according to any of the preceding claims, wherein the segmentation analysis comprises the processing of the vectorized attribute data using one or more segmentation algorithms.

10. A method according to claim 9, wherein the segmentation analysis comprises the processing of the vectorized attribute data using multiple segmentation algorithms to produce multiple segmentation outcomes.

1 1. A method according to claim 10, where the multiple segmentation outcomes are compared.

12. A method according to any of the preceding claims, wherein machine learning is used in the segmentation analysis.

13. A method according to any of the preceding claims, wherein the method further comprises the step of: constructing a structural model of the subsurface location using the segmented geophysical seismic data.

14. A system for the segmentation of seismic data of a subsurface location, the system comprising: a controller; storage storing electronic program instructions for controlling the controller; and an input means; wherein the controller is operable, under control of the electronic program instructions, to: receive input via the input means, the input comprising seismic data; process the input to generate an output set of seismic textural attributes; process the output set of seismic textural attributes to generate vectorized attribute data; process the vectorized attribute data using a segmentation analysis to generate an output segmented seismic data; and communicate the output segmented seismic data.

15. A system according to claim 14, wherein the system comprises a display for displaying a user interface, the controller is operable, under control of the electronic program instructions, to communicate the output set of seismic textural attributes by displaying the output via the display.

16. A system according to claim 14 or 15, wherein the controller is operable, under control of the electronic program instructions, to allow a user to assign weightings to the set of seismic textural attributes.