WO2015199727A1

WO2015199727A1 - Finding faults in geological data

Info

Publication number: WO2015199727A1
Application number: PCT/US2014/044632
Authority: WO
Inventors: Gregory Stuart Snider
Original assignee: Hewlett Packard Development Co LP
Current assignee: Hewlett Packard Development Co LP
Priority date: 2014-06-27
Filing date: 2014-06-27
Publication date: 2015-12-30
Anticipated expiration: 2016-12-27

Abstract

Finding faults in geological data is disclosed. One example is a system where input geological image is received. A semblance analyzer generates a semblance metric at each point of the image, the semblance metric being indicative of local semblance analysis. A tensor field extractor extracts a structure tensor at each point of the image, the structure tensor being indicative of local structure in the image. A phase congruency filter provides a stick tensor at each point of the image, the stick tensor based on the semblance metric, and being indicative of a point lying on a major edge in the image. An evaluator evaluates, at each point of the image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor being indicative of a likelihood of each point lying along a potential fault in the image.

Description

FINDING FAULTS M GEOLOGICAL DATA

Background

{OOOiJ Geological data may be analyzed to detect a geological fault. A geoiogical fault is a crack in the earth's crust resulting from a displacement of one !andmass relative to another. Such analysis of geological data may be utilized in oil and gas exploration.

Brief Description of the Drawings j 02] Figure 1 is a functional block diagram illustrating one example of a system for finding faults in geological data.

fO003J Figure 2 is another functional block diagram illustrating one example of a system for finding faults in geological data.

|lRi04| Figure 3 illustrates one example fo finding faults in geological data. j¾005j Figure 4 is a block diagram illustrating one example of a processing system for finding faults in geologicai data.

0006) Figure 5 is a block diagram illustrating one example of a computer readable medium for finding faults in geological data.

10007) Figure 8 is a flow diagram illustrating one example of a method for finding faults in geological data.

Detailed Description

{0008} Oil and gas exploration utilizes analysis of selsmological data to detect a potential geological fault. Typically, the amount of data extracted from a surve of a potential oii field is large, and a fraction of that data is analyzed, resulting in possible errors in fault detection. Local semblance analysis may be performed on geological data to detect a potential geological fault. Output of the semblance analysis may be noisy, and in some instances, the output may not clearly delineate a potential: geographical fault.

£0009 J As described in various examples herein, finding faults in geoiogicai data is disclosed. One example is a system that combines the semblance analysis with a filtering step to enhance data related to a potential geological fault. The example system comprises a semblance analyzer, a tensor field extractor, a phase congruence filter, and an evaluator. Input geological Image is received. The semblance analyzer generates a semblance metric at each point of the geological image, the semblance metric being indicative of local semblance analysis. The tensor field extractor extracts a structure tensor at each point of the geoiogicai image, the structure tensor being indicative of local structure in the geological image. The phase congruency filter provides a stick tensor at each point of the geological image, the stick tensor being based on the semblance metric, and being indicative of a point lying on a major edge in the geoiogicai image. The evaluator evaluates, at each point of the geological image, a fault tensor that is based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor being indicative of a likelihood of each point lying along a potential fault in the geological image. In particular, it interpolates fragmentary traces to provide smooth curves indicating potential faults.

( 0101 in the following detailed description, reference is made to the

accompanying drawings which form a part hereof, and in which is shown by way of illustration specific examples in which the disclosure may be practiced, it is to be understood that other examples may be utilized, and structural or logical changes may be made without departing from the scope of the present disclosure. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present disclosure is defined by the appended claims. It is to be understood that features of the various examples described herein may be combined, i part or whole, with each other, unless specifically noted otherwise.

£O0i l| Figure 1 is a functional block diagram illustrating one example of a system 100 for finding faults in geological data. The system 100 receives input geological data via a processing system, A processing system is a combination of machines that may be utilized to receive an input, process the input, and generate an output, in one exampie, the input geoiogicai data is processed by a sembiance analyzer and a tensor field extractor included in system 100, The semblance anaiyzer generates a sembiance metric at each point of the geoiogicai image, the sembiance metric indicative of local sembiance analysis. The tensor field extractor extracts a structure tensor at each point of the geoiogicai image, the structure tensor indicative of locai structure in the geoiogicai image. The semblance metric is processed by a phase congruency filter to provide a stick tensor at each point of the geoiogicai image, the stick tensor indicative of a point tying on a major edge in the geoiogicai image. The sembiance metric, structure tensor, and the stick tensor are processed by an evaluator to determine a fauit tensor, the fault tensor indicative of a likelihood of each point lying along a potential fault in the geoiogicai image,

f 0012 J System 100 includes input geological data 102, a semblance anaiyzer 104, a tensor field extractor 108, a phase congruency fitter 108, a stick tensor 110, and an evaluator 112. in one exampie, input geoiogicai data 102 is a geological image representing data related to geoiogicai strata, in one example, input geological data 102 is a geological image representing

seismological data. Such data may be received, for exampie, from a variety of data acquisition systems. In one exampie, the input geoiogicai data 102 may be a three dimensional image.

jOODj The sembiance anaiyzer 104 performs semblance analysis on the input geoiogicai data 102 to generate a sembiance metric at each point of the geoiogicai data. The sembiance metric is a quantitative measure of coherence of the geoiogicai data. The sembiance anaiyzer 104 computes local sembiance images using local smoothing filters. Local sembiance is a squared smoothed-image divided by a smoothed squa red-image, where smoothing is performed by local smoothing filters along the eigenvectors of a structure tensor.

i)l4j The tensor field extractor 106 extracts a structure tensor, The structure tensor is a structure tensor field, with a structure tensor associated wit each point in the geological data 102. In one example, the structure tensor is extracted via a technique for extracting local structure in images which produces a tensor field, with a 2 x 2 tensor produced for each point in the input geological data 02. In one example, the tensor field extractor 108 further extracts, for each point of the geological image, a smallest eigenvalue of the structure tensor. For example, the tensor field extractor 106 extracts, for each tensor in the structure tensor field, the ballness, the smallest eigenvalue of the structure tensor. BaS!ness is a measure of orientation uncertainty or isotropy. In one example, it is indicative of points of low coherence, thereby supplementing the semblance analysis with additional evidence.

[0915] System 100 includes a phase congruency filter 108. The semblance metric is passed through the phase congruency filter 108. in one example, the phase congruency filter 108 extracts edges In images, and may be adjusted to extract longer edges while filtering out shorter edges. This may be utilized to emphasize long traces in the semblance metric while suppressing short traces. The phase congruency filter 108 may be tuned to adjust parameters that are indicative of a length of a trace. Accordingly, traces that are longer than a threshold length may be extracted and retained, whereas traces that are shorter than the threshold length may be suppressed. The phase congruency filter 108 provides a field of stick tensors 110, a 2 x 2 tensor produced for each point in the Input geological data 102, where each tensor has a single non-zero eigenvalue and carries both orientation and certainty of the edges detected by the phase congruency filter 108.

[0016} The evaiuator 112 determines, at each point of the geological image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor indicative of a likelihood of each point lying along a potential fault in the geological image.

{0017} In one example, a magnitude of a product of the smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the semblance metric from unity is determined. In one example, the fault tensor may be determined as the following tensor field;

bailness x ( 1— semblance metric) x Stick Tensor, where ballness is the smallest eigenvalue of the structure tensor extracted by the tensor field extractor 106, the semblance metric is generated by the semblance analyzer 104, and the stick tensor 110 is provided by the phase congruency filter 108. The result is a field of sparse stick tensors that have a high likelihood of lying along a potential fault.

{6018} in one example, evaiuator 112 further interpolates the fault tensor to generate a continuous trace indicative of the potential fault in the geological image. For example, system 100 includes tensor voting which interpolates the fault tensor determined by the evaiuator 112, to create Song, smooth traces that follow Gestait principles of good continuation. The tensor voting fills in discontinuities in the fault tensor based on an assumption of continuity of a geological fault line.

{0019} In one example, the evaSuator further sharpens the continuous trace. For example, system 100 includes non-maximum suppression to sharpen the fault tensor. Generally, the output of tensor voting may be a blurred image. Non- maximum suppression provides a sparser, but contour-continuous fault tensor. |lKi20| In one example, the evaiuator 112 further determines a magnitude of the fault tensor at each point of the geological image. For example, the largest eigenvalue of the fault tensor may be computed for each point of the geological data.

( 0211 in one example, system 00 may include a highlighter to superimpose the potential fauit in the geological image onto the input geological image, in one example, an enhanced fault tensor may be generated based on tensor voting and/or non-maximum suppression, and the enhanced fault tensor may be superimposed on an input geological image to generate an output Image. Thus, the potential geological faults in the input geological image are highlighted. Accordingly, the noise typical of a semblance analysis is filtered out and the potential geological fault is highlighted.

|0022) Figure 2 is another functional block diagram illustrating one example of a system for finding faults in geological data. Input geological data 202 is received. Semblance analysts 204 is performed on the input geological data 202 to generate a semblance metric at each point of the geological data 202, trie sembiance metric indiicattve of iocai sembiance analysis. A structure tensor 208 is extracted at each point of the geologicai data 202, the structure tensor 208 indicative of iocai structure in the geologicai data 202. The semblance metric is filtered via a phase congruency 208 to provide a stick tensor at each point of the geological data 202, the stick tensor based on the semblance metric, and indicative of a point lying on a major edge represented by the geologicai data 202.

[0023 j The sembiance metric, the stick tensor, and the structure tensor are combined to determine a fault tensor 210. In one example, a magnitude of a product of the smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the semblance metric from unity is determined.

|0i>24] In one example, the fault tensor 210 is further refined by applying tensor voting 212, in one example, the tensor resulting from the tensor voting 212 is furthe refined by applying non-maximum suppression 214, In one example, the fault tensor 210 is further refined by applying the non-maximum suppression 214, In one example, a magnitude of the fault tensor 210 is computed to provide an enhanced fault tensor 216. for example, the largest eigenvalue of the fault tensor may be computed for each point of the geological data 202, In one example, the enhanced fault tensor 218 is based on th tensor resulting from the tensor voting 212. In one example, the enhanced fault tensor 216 is based on the tensor resulting from the non-maximum suppression 214,

[0025] Figure 3 illustrates one example for finding faults in geologicai data, input geological image 302 is received. Semblance analysis 304 is performed on the input geologicai image 302 to generate a semblance metric at each point of the geological image 302, the semblance metric indicative of Iocai semblance analysis 304. A structure tensor is extracted at each point of the geological image 302, the structure tensor indicative of local structure in the geological image 302, In one example, the structure tensor is the ballness 306, the smallest eigenvalu of the structure tensor 306.

[0026] In one example, the semblance metric from semblance analysis 304 and the bailness 306 are combined to provide the fault tensor. In one example, a magnitude 308 of the fault tensor is computed. In one exampie, an enhanced fault tensor 310 is based on the tensor resulting from applying tensor voting to the fault tensor, in one example, the enhanced fault tensor 310 is

superimposed on the input geological image 302 to generate output image 312. As illustrated, a first fault line 314 and a second fault line 316 may be

superimposed onto the input geological image 302 to generate the output image 312.

{00271 Figure 4 is a block diagram illustrating one example of a processing system 400 for implementing the system 100 for finding faults in geological data. Processing system 400 includes a processor 402, a memory 404, input devices 416, and output devices 418. Processor 402, memory 404, input devices 416, and output devices 418 are coupled to each other through communication link {e.g., a bus).

i&02$3 Processor 402 includes a Central Processing Unit (CPU) or another suitable processor, in one example, memory 404 stores machine readable instructions executed by processor 402 for operating processing system 400. Memory 404 includes any suitable combination of volatile and/or non-volatile memory, such as combinations of Random Access Memory (RAM), Read-Only Memory (ROM), flash memory, and/or other suitable memory.

fftfl29j Memory 404 stores input geological data 406 for processing by processing system 400. In one example, memor 404 stores a seistrsoiogicaS image for processing by processing system 400. In one example, memory 404 stores a three-dimensional image for processing by processing system 400. Memory 404 also stores instructions to be executed by processor 402 including instructions for a semblance analyzer 408, a tensor field extractor 410, a phase congruency filter 412, and an evaluate* 414. In one example, semblance analyzer 408, tensor field extractor 4 0, phase congruency filter 412, and evaiuator 414, include semblance analyzer 104, tensor field extractor 106, phase congruenc filter 108, and ©valuator 1 12, respectively, as previously described and illustrated with reference to Figure 1. In one example, memory 404 stores semblance metric, structure tensor, stick tensor, and fault tensor for processing by processing system 400, {0030} in one example, processor 402 utilizes the input geologica! data 408 and executes instructions of semblance analyzer 408 to generate a semblance metric at each point of the geological image, the semblance metric being indicative of local semblance analysis. Processor 402 also utilizes the input geological data 408 and executes instructions of tensor field extractor 410 to extract a structure tensor at each point of the geological image, the structure tensor being indicative of local structure in the geological image.

1003 J In one example, the processor 402 utilizes the semblance metric and executes instructions of phase congruency filter 412 to provide a stick tensor at each point of the geological image, the stick tensor being indicative of a point lying on a major edge in the geoiogicai image, !n one example, processor 402 executes instructions of joint edge-aware filter 412 to provide piecewise smoothing of the input data based on a spatial measure of each data element in the input data, where the spatial measure is indicative of a maximal spatial extent of the piecewise smoothing at each data element. In one example, processor 402 executes instructions of evaiuator 414 to determine, at each point of the geological image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor being indicative of a likelihood of each point lying along a potential fauit in the geological image. 10032} input devices 416 include a keyboard, mouse, data ports, and/or other suitable devices for inputting information into processing system 400. In one example, input devices 416 are used to input geological data 406, such as a geological image. Output devices 418 include a monitor, speakers, data ports, and/or other suitable devices for outputting information from processing system 400. As described herein, in one example, an enhanced fault tensor may be generated based on tensor voting and/or non-maximum suppression, and the enhanced fault tenso may be superimposed on the input geological image to generate an output image. Output devices 418 are used to output the output image with the potential geoiogicai fault highlighted.

[0033) Figure 6 is a block diagram; illustrating one example of a computer readable medium for finding faults in geological data. Processing system 500 includes a processor 502, a computer readable medium 510, a semblance analyzer 504, a tensor field extractor 508_s and a phase congruency filter 508. Processor 502, computer readable medium 510, the semblance analyzer 504, the tensor field extractor 506, and the phase congruency filter 508 are coupled to each other through communication link (e.g., a bus).

(0034] Processor 502 executes instructions included in the computer readable medium 510, Computer readable medium 510 includes geological data receipt instructions 512 to receive input geoiogicaf image. Computer readable medium 510 includes semblance metric generation instructions 514 of the semblance analyzer 504 to generate a semblance metric at each point of the geological image, the semblance metric being indicative of local semblance analysis.

5] Computer readable medium 510 Includes structure tensor extraction instructions 518 of the tensor field extractor 506 to extract a structure tensor at each point of the geological image, the structure tensor being indicative of local structure in the geological image. Computer readable medium 510 includes stick tensor provision instructions 518 of the phase congruency filter 508 to provide a stick tensor at each point of the geologica! image, the stick tensor being based on the semblance metric, and being indicative of a point lying on a major edge in the geological image. Computer readable medium 510 includes fault tensor evaluation instructions 520 to evaluate, at each point of the geological image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor being indicative of a likelihood of each point lying along a potential fault in the geological image. Computer readable medium 5 0 includes fault superimposiiion instructions 522 to superimpose the potential fault In the geological image onto the input geological image,

{0036] In one example, computer readable medium 510 includes further instructions to evaluate, at each point, a magnitude of a product of a smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the semblance metric from unity.

[0037) In one example, computer readable medium 510 includes further instructions to provide the input geological image with the superimposed potential fault. |0 38) Figure 8 is a Fiow diagram illustrating one example of a method for finding faults in geological data. At 600, a geological image is received. At 602, a semblance metric is generated at each point of the geological image. At 604, a structure tensor is extracted at each point of the geological image. At 606, a stick tensor is provided at each point of the geological image, the stick tensor being based on the semblance metric. At 608, a fault tensor is evaluated based on the semblance metric, the structure tensor, and the stick tensor. At 610, the fault tensor in the geological image is superimposed onto the input geological image.

)3 j In one example, the method includes providing the input geological image with the superimposed potentiai fault.

{ΘΘ40] In one example, extracting the structure tensor includes extracting, for each point of the geological image, a smallest eigenvalue of the structure tensor. In one example, evaluating the fault tensor includes evaluating a magnitude of a product of the smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the semblance metric from unity.

jlRWI I In one example, evaluating the fault tensor includes interpolating the fault tensor to generate a continuous trace indicative of the potentiai fault in the geological image. In one example, evaluating the fault tensor Includes sharpening the continuous trace.

( 21 in one example, evaluating the fault tensor includes determining a magnitude of the fault tensor at each point of the geological image.

10043 j Examples of the disclosure provide a generalized system for finding faults in geologicai data based on semblance analysis enhanced with a filtering step to identify data related to a potential geologicai fault. The generalized system provides a filter-based, automatable approach to finding potential fault in geological data by processing large geological datasets and looking for patterns that have a high likelihood of being potentiai faults, and filtering out other patterns. The generalized system extracts long, potential fault structures while suppressing short structures. The disclosed processes are parallelizable to run on multicore platforms, such as graphics processing units ("GPUs"). The disclosed processes are applicable to 3-D datasets. |0044] Although specific examples have been sliusirated and described herein, a variety of alternate and/or equivalent implementations ma be substituted for the specific examples shown and described without departing from the scope of the present disclosure. This application is intended to cover any adaptations or variations of the specific examples discussed herein. Therefore, it is intended that this disclosure be limited only by the claims and the equivalents thereof.

Claims

1 , A sysiem comprising:

input geological image received via a processing system; a semblance analyzer to generate a semblance metric at each point of the geological image, the semblance metric indicative of local semblance analysis;

a tensor field extractor to extract a structure tensor at each point of the geological image, the structure tensor indicative of local structure in the geological image;

a phase congruency filter to provide a stick tensor at each point of the geological image, the stick tensor based on the semblance metric, and indicative of a point lying on a major edge in the geological image; and

a evaluator to determine, at each point of the geological image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor indicative of a likelihood of each point iying along a potential fault in the geological image.

2. The system of claim 1 , wherein the tensor field extractor further extracts, for each point of the geological image, a smallest eigenvalue of the structure tensor.

3, The system of claim 2, wherein the evaluator further determines a

magnitude of a product of the smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the semblance metric from unity.

4. The system of claim 1 , wherein the evaluator further interpolates the fault tensor to generate a continuous trace indicative of the potential fault in the geological image.

5. The system of claim 4, wherein the evaiuator further sharpens the continuous trace.

6. The system of c!aim 1 , wherein the evaiuator further determines a

magnitude of the fault tensor at each point of the geological image.

7. The system of ciaim 1 , further comprising a highlighter to superimpose the potential fault in the geologicai image onto the input geological image,

8. A method to determine a potential fault in geologicai data, the method comprising;

receiving, via a processor, a geologicai image;

generating a semblance metric at each point of the geological image, the semblance metric indicative of local semblance analysis; extracting a structure tensor at each point of the geological image, the structure tensor indicative of local structure in the geological image; providing a stick tensor at each point of the geological image, the stick tensor based on the semblance metric and indicaiive of a point lying on a major edge in the geologicai image;

evaluating, at each point of the geological image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor indicative of a likelihood of each point lying along the potential fault in the geologicai image; and

superimposing the potential fault in the geological image onto the input geologicai image.

9. The method of claim 8, further comprising providing the input geologicai image with the superimposed potentiai fault.

10. The method of claim 8, wherein evaluating the fault tensor further

comprises evaluating a magnitude of a product of a smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the semblance metric from unity.

1 .The method of claim 8, wherein evaluating the fault tensor further

comprises interpolating the fault tensor to generate a continuous trace indicative of the potential fault in the geological image.

12. The method of claim 1 1 , wherein evaluating the fault tensor further

comprises sharpening the continuous trace.

13. The method of claim 8, evaluating the fault tensor further comprises determining a magnitude of the fault tensor at each point of the geoiogicai image.

14. A non-transitory computer readable medium comprising executable

instructions to:

receive, via a processor, a geoiogicai image;

generate a semblance metric at each point of the geoiogicai image, the semblance metric indicative of local semblance analysis; extract a structure tensor at each point of the geological image, the structure tensor indicative of local structure in the geoiogicai image; provide a stick tensor at each point of the geological Image, the stick tensor based on the semblance metric and indicative of a point lying on a major edge in the geoiogicai image;

evaluate, at each point of the geological image, a fault tensor based on the semblance metric, the structure tensor, and the stick tensor, the fault tensor indicative of a likelihood of each point lying along a potential fault in the geological image; and

superimpose the potential fault in the geoiogicai image onto the input geoiogicai image.

15. The non-transitory computer readable medium of claim 14, wherein the executable instructions include further instructions to evaluate, at each point, a magnitude of a product of a smallest eigenvalue of the structure tensor, the stick tensor, and the difference of the sembiance metric from unity.