Sei sulla pagina 1di 13

An Approach towards Automated Fault Interpretations in Seismic Data

Fitsum Admasu Otto-von-Guericke Universit at Magdeburg Klaus T onnies Otto-von-Guericke Universit at Magdeburg

Abstract In this paper, we present a methodology for interpreting faults from three dimensional seismic data. Faults are individual fractures across which there are visible offsets of horizons (or rock layers). 3D seismic data - images of subsurface structure generated by reecting seismic waves off rock layers - have been used for hypothesizing subsurface structures. Since interpretation of seismic data is a highly time-consuming task, automated tools to assist the interpretation are crucial. Our work focuses on automating the correlation of horizons across a fault so that helping in dening the faults geometry. The correlation is made by integrating empirical structural geological models into normalized cross-correlation. We employ a multi-resolution approach dened on perceptual scale. Though still detailed evaluations are required, the results show correct matches. In areas of weaker signals, or where the seismic data are less clear, the results are incorrect correlations.

1 Introduction
When seismic waves are sent to underground structures, their velocities change due to different acoustic impedances of subsurfaces rock layers. These changes in velocity result in reections which are recorded by sensors on the surfaces and appear on the seismic images. The seismic images usually come as 3d recording of subsurface cross-section and are considered as sequence of slices [Dor98]. The strong horizontally layered reection events visible on the seismic images are known as horizons and represent underground rock layers. A fault surface forms discontinuity in the rock, where rock on either side of the fault is displaced relative to the rock on the opposite side. Layers of rock which are observed on the seismic data and that have been moved by the action of faults are called faulted horizons. Unless erosion occurred, the faulted horizons usually have their corresponding part on the other side of the fault. The faulted horizons offset is maximum at the mid of the fault and decreases to zero towards the tips of the fault [WW88]. The correspondence analysis between faulted horizons across a fault, that is nding the offsets of these faulted horizons,
Fakult at Fakult at

f ur Informatik, Institut f ur Simulation and Graphik, D-39016 Magdeburg, Germany f ur Informatik, Institut f ur Simulation and Graphik, D-39016 Magdeburg, Germany

is important for describing the fault. Accurate assessment of fault geometry and displacements are of particular importance in planning the most efcient way to extract oil and gas from underground. Thus, the correlation of horizons across faults is an indispensable task of seismic interpretation. Seismic data interpreters locate faults as lines from horizon discontinuities on seismic slices. Then they connect horizon segments across faults on the basis of reection character and geological reasoning. Since the human eyes are restricted within a two-dimensional section, the interpreters evaluate their correlation decision by using the 2-d projections of the 3-d seismic data. Interpretation of some geological features are done manually for a seismic slice shown on gure 1. However identication of these geological features in seismic sections by an interpreter is time consuming and subjective.

Horizons

Fault lines Correlated horizons

Figure 1: Seismic slice with manually interpreted faults and horizons. The main focus of this research work is developing a computer-based methodology for correlation of horizons across faults. Expected outcomes of this automation are reducing the time-consuming manual task, and avoiding the uncertainties associated with fault interpretation by providing a repeatable and robust seismic data analysis tool.

Depth

2 Previous Works
Some automated tools have been developed to assist interpretation of horizons and fault surfaces of seismic data. The commonly used ones are auto-picking or auto-trackers (reviewed in [Aur03], [Dor98]). Auto-picking tools are aimed at extending manually selected seismic traces based on local similarity measures. They perform well if there are uninterrupted horizon features. But horizon discontinuities are very common. Alberts et.al. [WL00] explain a method for tracking horizons across discontinuities. They trained articial neural networks (ANN) to track similar seismic intensity. However, horizon tracking across faults using solely seismic patterns is infeasible due to large seismic data distortion near faults. To alleviate this matter, Aurnhammer and Toennies [AT02] propose a model-based scheme for matching horizons at normal faults in 2D seismic images. Well-dened horizons segments on both sides of the fault were extracted and matched based on local correlation of seismic intensity and geological knowledge. Since exhaustive search for optimal solution of correlation is unfeasible, genetic algorithm as optimization technique was utilized. However, a pure two-dimensional approach lacks efciency and is suitable only if the information of the 2D seismic slice is sufcient for evaluation of the geological constraints. Previous work in our group [AT04] describes a multi-resolution continuous horizon correlation scheme where the correlation task is formulated as a non-rigid continuous point matching between the two sides of the fault. Continuous means each point on the left side of the fault is assigned a corresponding position on the right side. The continuous point matching approach has the advantage that it does not require all horizons to be welldened. However, it is computational expensive and not sufciently robust with respect to noise and artifacts in seismic data. Besides, interactions from nearby faults distort the global fault displacement model which was computed at the very coarse level. The human interpreters usually extract signicant horizons on either sides of the fault line on the 2d seismic slices and propagate to the subsequent slices to identify if there is a strong feature such as zero offset of the fault. If there is such feature then they return to the previous slices tracking the fault offset and identify the offsets of the horizons at the high fault offset regions. Then they go to the less prominent horizons and try to nd the correspondences. These interpreters practices are the basis for our matching model here. Horizon segments are extracted on the fault surface, and then matching between these segments is done by nding strong horizons signals which give guidance for matching the weaker horizons signals.

3 Method
We have designed a matching priori which constitutes a seismic based normalized crosscorrelation model and a geological fault displacement model. Then, correlation of horizons across faults is carried out in four steps: 1. Fault Patch Computations 2. Feature Computations

3. A-prior Matching Model 4. Optimal Solution Search These steps are described in the following sections.

3.1 Fault Patch Computations


Usually seismic data consist of large numbers of faults and fault systems; however, we restrict ourself to correspondence analysis localized near fault regions and to a single fault surface. As result we need a tool to extract a fault patch, a subset of the seismic data, which contains only a fault surface with uninterrupted seismic sections on the two sides of the surface. A fault is a 3d damage zone on the layers of horizons. A fault surface is a regression surface that ts the damage zone. Various methods for automatic fault surface extraction from seismic data have been reported in publications by Steen et. al.[ML01], Bahorich et.al. [BF95] and Gibson et.al [ST03]. However, these methods are not yet fully used by geologists due to their limited success. Developing a fully automatic fault extraction method is very challenging due to the complicated geometry of faults and seismic noises which easily misguide any fault tracking tools. We have developed a technique to extract the fault surface semi-automatically. An operator provides the initial fault tracking direction as well as corrections in areas of low signal to noise ratio. A fault surface is extracted from discontinuities of horizons in seismic data. Thus the rst step of any fault extraction algorithm is to highlight such discontinuities in the seismic data. Different techniques such as coherence cube [BF95], semblance [FB98], the eigenstructure of the data covariance matrix [GM99] are already introduced for discontinuity enhancing in seismic data. These techniques are designed to enhance spatial discontinuities computed at every point. They are very sensitive for random structures and known for high timecomplexity. For our fault extraction tool, we found the lter response of the log-Gabor lter appropriate. The log-Gabor lter is less sensitive to random structures and faster. The orientation selectivity of the lter provides linear-like structure features which are more suitable for fault modelling. The seismic image is convolved with a set of log-Gabor lters at different orientations and different scales. This technique has already been successfully applied for digital image partition and boundary detection [FVFV99]. Then tracking of the fault surface is performed on the lter response of log-Gabor lter. A fault line on a slice is given by a user-specied line. Then the automatic extraction is done by propagating and identifying the fault lines in the successive slices using linear regression and orientation constraints. Later a fault surface is constructed by spline interpolation between these lines. The fault lines generation steps are shown on gure 2.

3.2 Feature Computation


The fault surface which is extracted from the previous fault tracking is used as input here. From each side of the fault surface (or plane), local features are projected along the horizons orientations onto the fault plane. Seismic information is distorted at locations close

(a)

(b)

(c)

(d)

Figure 2: (a) Seismic slice. (b). Fault enhancing by log-Gabor lters. (c). Automatically detected fault line on the post-processed image of (b). (d) Automatically detected fault line on the original seismic slice. to the fault because of the geological process of fault creation. To correct for this fault distortion, averaging along the horizon starts at some distance from the fault line. Then features are projected to the fault lines. The orientations of the horizons are dened by the Canny edge detector [Can86]. In the cases where the seismic data are already anisotropically ltered [Bak02], we take the pixels values at ve pixel distance from the fault as mapped feature to the fault plane. These processes produce left and right fault feature 2D images (see Figure 3). Figure 4 and 5 show projected feature images from different fault patches. Each column of these images shows seismic features projected to the fault line on the seismic slice at that vertical position. The feature mapping process produces two feature planes from a fault patch; henceforth we call these features as left and right fault planes. Furthermore, horizon segments are extracted from these two fault planes. The horizon segments are extracted by taking the peak values on the column of each feature plane (see gure 6). The following features are computed for each segment: Complex seismic attributes (amplitude, phase) around a neighbor window. Strength of reection - this is a relative measure obtained by comparing the contrast of the seismic amplitude around the segments. Position - the depth of the segment in relative to other segments.

3.3 A-prior Matching Model


The correlation problem can be seen as a registration or stereo correspondence problem between the two feature planes. However, the application of classical stereo correspondence

Fault Plane

Left side of the fault mapped to the left plane

Right side of the fault mapped to the right plane

Horizons

Fault patch

Mapping from fault patch to fault plane

Figure 3: A fault patch is mapped onto left and right fault planes. [SZ01] or registration algorithms [MV98][Bro92] to perform the correspondences between the feature images is not feasible due to little intensity information to guide the utilization of optical ows and presence of local distortion. We dene the horizon correlation as a labelling problem in which the segments, L, extracted from the left feature image serve as sites and the right side segments, R, serve as labels, and it can be paralleled with MAP-MRF (Maximum a posteriori-Markov Random Field) framework advocated by Geman and Geman [GG84]. The labelling function, , is dened as : L R {} (1) where represents not-segment regions and is assigned to sites where there are no corresponding labels from R. Each site is considered as a random variable, and the labelling as events. When all the sites have some labelling assigned to them we have a conguration, henceforth denoted as T . However, the admissible labels may not be common to all the sites due to the geological constraints that horizons must not cross each other. Furthermore, as we deal with only normal faults, where the hanging wall moves down relative to the footwall, offsets have only one direction. These impose constraints on the search for wanted congurations. As it was pointed by previous publication [AT04][Aur03], we can not rely only on the seismic information to solve the correlation task. The correlation task needs to be guided by a priori knowledge of displacement patterns on the fault surface. Therefore, we need an ob-

Figure 4: Seismic feature planes representing the left and right side of the fault surface projected from unltered dataset. jective function which maps a candidate conguration solution to a real number measuring the quality of the solution in terms of seismic similarity as well as geological knowledge. Such objective function is dened as follows. (T ) = Es (T ) + Eg (T ) + Ec (T ) (2)

where Es is computed as the normalized cross-correlation coefcient value between seismic (amplitude and phase) features of candidate segments pairs, here modelled as sites and labels. The normalized cross-correlation technique has been already successfully used before by Aurnhammer [Aur03] to measure the similarity between seismic signals. Its strength comes from its ability to measure linear relationships of the seismic features. Eg and Ec measure the interaction potentials between labels of the sites. Ec is MRF-based smoothness constraint while Eg derived from fault displacement model explained in the next section. 3.3.1 Fault Displacement Model

According to heuristics of Walsh et.al. [WW87], a normalized displacement, D, at a point on a fault surface is given by D = 2(((1 + r)/2)2 r2 )2 (1 r) (3)

where r is the normalized radial distance from the fault center. The normalized displaced ment is D = dmax where d is the fault displacement at a point and dmax is the maximum displacement on a fault surface. Then Eg at equation 2 is computed as the least square error between a given current conguration offset and the theoretical transformation map at equation 3.

Figure 5: Seismic feature planes representing the left and right side of the fault surface projected from anisotropically ltered dataset.

3.4 Optimal Solution Search


The solution for the horizon correlation posed as labelling problem at equation 1 is a conguration, Tmax , which maximizes the value of the objective function described at equation 2. Geman and Geman [GG84] provides a proof for such claim, assuming Markov Random Field distribution. Since searching for Tmax is not trivial due to the non-linearity and many local maxima, we use a simulated annealing (SA) [GV83], a stochastic non-linear search optimization technique. Some horizons segments on the left may not have corresponding segments on the right side. We handle such cases by dening a local similarity function for such missed segments based on interpolated offsets from well-dened horizons segments. The search for Tmax uses a perception-based multi resolution framework where horizons signals with higher strength of reection guide the matching at horizon signals with lesser strength of reection. The horizon segment matching algorithm is illustrated at algorithm 1.

Figure 6: Horizon segments generated on the feature planes. The width of line indicates the strength of the reection and so denes the resolution level, the wider the coarser.

Data: LeftSeg, RightSeg Result: MatchPair Function MatchPair = SegMatch(LeftSeg, RightSeg); if LeftSeg is empty or RightSeg is empty then return []; else RightSegH = selectStrongReection(RightSeg); LeftSegH = selectStrongReection(LeftSeg); MatchPair = SimAnnealing(LeftSegH,RightSegH); Partition = partiton(LeftSeg,RightSeg,MatchPair); for each Partition(i) do MatchPair=[MatchPair, SegMatch(Partition(i).Left, Partition(i).Right)]; end end Algorithm 1: Recursive segment-matching algorithm.

4 Experiments and Results


Our experimental data consist of several fault patches taken from shallow regions of real 3D seismic data. The fault patches were extracted semi-automatically using the method described in section 3.1 . Each of these fault patches contains a normal fault and has at least three well-dened horizons. The left and the right side horizon segments were extracted using the technique described in section 3.2. Then the correlations between these segments were computed using the segment-matching algorithm illustrated at algorithm 1. Some of the correlations results are demonstrated on gures 7 - 9. The results are given on seismic slices restored from the feature planes of different faults. The original seismic slices and the automatic correlation results for prominent horizons are shown. While using solely the seismic intensity information for the matching criteria we were not able to get correct correlations for any test cases. However for fault 1 shown on gure 7, we were able to obtain correct correlations without applying the geological fault displacement model, which means using only the smoothness constraints and the local intensity information. This appears to be due to relatively small size and offsets of the fault and similar intensity and spatial proles of the horizons at both sides of the fault. For faults 2 (on gure 7) and faults 3 and 4 (on gure 8), the constraint from fault displacement model described in section 3.3.1 was necessary to arrive at the correct correlations. However, for faults 3 and 4, we were not able to obtain the correct matches using the continuous matching algorithm described in [AT04]. The segment-matching algorithm was not successful for faults 5 and 6 shown on gure 9. These failures are mainly due to local features disturbances which are also partially resulted from incorrect denitions of the fault surface.

Fault 1

Fault 2

(a)

(b)

(c)

(d)

Figure 7: Automatic correlation results (black arrows) for some prominent horizons. (a) and (b) show the original seismic slice and the correlation results for fault 1. (c) and (d) do the same for fault 2.

Fault 3

Fault 4

(a)

(b)

(c)

(d)

Figure 8: Automatic correlation results for some prominent horizons. (a) and (b) show the original seismic slice and the correlation results for fault 3. (c) and (d) do the same for fault 4.
Fault 5 Fault 6

(a)

(b)

(c)

(d)

Figure 9: Black arrows show the automatic correlation results. (a) and (b) show the original seismic slice and the correlation results for fault 5. (c) and (d) do the same for fault 6. For some incorrect results, the white arrows show the manual correlations.

5 Discussions
The validity of the results is actually a subjective decision and more evaluations are necessary. Exception of the subjective decisions are cases where we know for sure the solution for the correlation. Such cases are when the fault terminates in the seismic data, we have zero offset regions of the faulted horizons. Then automatic interpretation is conrmed using closed loop that circumscribes interpreted fault at each horizon level. Matching sequences between the two feature planes was more successful than continuous matching described in our previous work [AT04]. With the change from isotropic features of the real part of the signal to anisotropic features of the complex signal, we increased the discriminative power for the local matching attribute. However, the usefulness for providing a reliability measure has to be determined. At very noisy regions of the seismic data, the cross-correlation coefcient is not reliable enough to estimate the seismic similarity; thus most of the incorrect

correlations are obtained for weaker horizon signals at deeper locations. The tool used currently for generating the fault surface produces 2D lines on each slice and doesnt merge them to create a smooth surface. This contributes for discontinuities. Another reason is the simple thinning algorithm used here is not able to extract the horizon segments everywhere due to noise artefact. Thus a more robust similarity measure derived by computing the internal conguration (texture) attributes of the regions is required. More studies regarding the stochastic optimization are necessary because the current optimization parameters are more in the nature of experience than specic guidelines; the relations of the parameters with the multi-resolution also need further investigations. The schedule for our optimization using simulated annealing actually is a simulated quenching process. It is faster than using the required schedule but, being a heuristic, may end in unwanted local minima. Parameters, such as initial condition and temperature schedule need to be found, given our sequence of optimizations from the multi-resolution representation.

6 Acknowledgements
We would like to acknowledge Shell for the seismic data and stimulating discussions. This research is supported by DFG Grant TO-166/8-1.

References
[AT02] M. Aurnhammer and K. Toennies. Model-based analysis of geological structures in seismic images. In: Simulation und Visualisierung, pp. 315327. SCS European Publishing House, March 2002. F. Admasu and K. Toennies. A model-based approach to automatic 3d seismic horizons correlations across faults. In: Simulation und Visualisierung, pp. 239 250. SCS European Publishing House, March 2004. M. Aurnhammer. Model-based Image Analysis for Automated Horizon Correlation across Faults in Seismic Data. PhD thesis,University of Magdeburg, Magdeburg, 2003. P. Bakker. Image structure analysis for seismic interpretation. PhD thesis,Technical University of Delft, 2002. M. Bahorich and S. Farmer. 3-D seismic Discontinuity for faults and stratigraphic features. The leading edge, 14(10):10531058, October 1995. L. Brown. A Survey of Image Registration Techniques. ACM Computing Surveys, 24(4):325376, 1992. J. Canny. A computational approach to edge detection. IEEE Trans. patt. anal. mach. intell., 8(6):679698, 1986.

[AT04]

[Aur03]

[Bak02] [BF95] [Bro92] [Can86]

[Dor98] [FB98]

G. Dorn. Modern 3-D Seismic Interpretation. The Leading Edge, 9(7):1262 1273, 1998. K. Marfurt R. Kirlin S. Farmer and M.S. Bahorich. 3-D seismic attributes using a semblance-based coherency algorithm. Geophysics, 63:11501165, 1998.

[FVFV99] R. Rodriguez-Sanchez J.A. Garcia J. Fdez-Valdivia and Xose R. Fdez-Vidal. The RGFF Representational Model: A System for the Automatically Learned Partitioning of Visual Patterns in Digital Images. IEEE Trans. Pattern Analysis and Machine Intelligence, 21(10):10441073, 1999. [GG84] S. Geman and D. Geman. Stochastic relaxation, Gibbs distribution and the Bayesian restoration of images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 6(6):721741, 1984. A. Gersztenkorn and K. J. Marfurt. Eigenstructure-based coherence computations as an aid to 3-D structural and stratigraphic mapping. Geophysics, 64:14681479, 1999. S. Kirkpatrick J. Gelatt and M. Vecchi. Optimization by Simulated Annealing. Science, 220(4598):671680, 1983. K. Tingdahl . Steen P. Meldahl and J. Ligtenberg. Semi-automatic detection of faults in 3-D seismic signals. In: 71st Annual Meeting, Expanded Abstracts, pp. 19531956, San Antonio, 2001. J. Maintz and M. Viergever. A survey of medical image registration. Journal of Medical Image Analysis, 2(1):136, 1998. D. Gibson M. Spann and J. Turner. Automatic Fault Detection for 3D Seismic Data. In: 7th Digital Image Computing: Techniques and Applications, pp. 821 830, Sydney, December 2003. D. Scharstein R. Szeliski and R. Zabih. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. In: IEEE Workshop on Stereo and Multi-Baseline Vision, Kauai, HI, December 2001. P. Alberts M. Warner and D. Lister. Articial Neural Networks for Simultaneous Multi Horizon Tracking across Discontinuities. In: 70th Annual International Meeting, Expanded Abstracts, pp. 239250, Calgary, Canada, March 2000. J. Walsh and J. Watterson. Distributions of cumulative displacement and seismic slip on a single normal fault surface. Journal of Structural Geology, 9(8):10391046, 1987. J. Walsh and J. Watterson. Analysis of the relationship between displacements and dimensions of faults. Journal of Structural Geology, 10(3):239247, 1988.

[GM99]

[GV83] [ML01]

[MV98] [ST03]

[SZ01]

[WL00]

[WW87]

[WW88]

Potrebbero piacerti anche