Satinder Chopra 1 and Kurt J. Marfurt 2. Search and Discovery Article # (2010) Posted June 28, Abstract

Transcription

1 AV Delineating Fractures from Seismic Attributes* Satinder Chopra 1 and Kurt J. Marfurt 2 Search and Discovery Article # (2010) Posted June 28, 2010 *Adapted from oral presentation at Session, Geophysical Integration: A Road Map to Exploration Success, at AAPG Annual Convention and Exhibition, New Orleans, April 11-14, Reservoir Services, Arcis Corporation, Calgary, AB, Canada (schopra@arcis.com) 2 ConocoPhillips School of Geology and Geophysics, University of Oklahoma, Norman, OK Abstract Fractures can enhance permeability in reservoirs and hence impact the productivity and recovery efficiency in those areas. Fold and fault geometries, stratal architecture and large-scale depositional elements (e.g., channels, incised valley-fill, and turbidite fan complexes) are often difficult to see clearly on vertical and horizontal slices through the seismic reflection data. Seismic attributes help us in characterizing stratigraphic features that may comprise reservoirs, and they form an integral part of most interpretation projects completed today. Coherence, curvature, and relative acoustic impedance are some important seismic attributes that are used for such analysis. However, for extracting accurate information from seismic attributes, the input seismic data needs to be conditioned optimally. This includes noise removal, using robust dip-steering options and superior algorithms for computation of seismic attributes. Curvature attributes in particular exhibit detailed patterns for fracture networks that can be correlated with image log and production data to ascertain their authenticity. One way to do this correlation is to manually pick the lineaments seen on the curvature displays for a localized area around the borehole, and then transform these lineaments into rose diagrams to compare with similar rose diagrams obtained from image logs. Favorable comparison of these rose diagrams lends confidence in the interpretation of fractures. Another way is to generate automated 3D rose diagrams from seismic attributes and correlate them with other lineaments seen on the coherence attribute. 3D volume rendering is one form of visualization that involves opacity control to view the features of interest inside the 3D volume. A judicious choice of opacity applied to edge-sensitive attribute sub-volumes, such as curvature or coherence co-rendered with the seismic amplitude volume, can both accelerate and lend confidence to the interpretation of complex structure and stratigraphy. Volume visualization of stratigraphic features is a great aid in 3D seismic interpretation and can be greatly aided by adopting cross-plotting of seismic discontinuity attributes in the interpretation workflow as we demonstrate in this presentation. Copyright AAPG. Serial rights given by author. For all other rights contact author directly.

2 References Al-Dossary, S. and K.J. Marfurt, 2006, 3D volumetric multispectral estimates of reflector curvature and rotation: Geophysics, v. 71/5, p. P41- P51. Chopra, S. and G. Larsen, 2000, Acquisition footprint it s detection and removal: CSEG Recorder, v. 25/8, p Gersztenkorn, A. and K.J. Marfurt, 1999, Eigenstructure-based coherence computations as an aid to 3-D structural and stratigraphic mapping: Geophysics, v. 64/5, p Hart, B.S., T. Engler, R. Pearson, and R.L. Robinson, 2002, 3-D seismic horizon-based approaches to fracture-swarm sweet spot: AAPG 2002 Annual Meeting Expanded Abstracts, p. 73. Lisle, R.J., 1994, Detection of zones of abnormal strains in structures using Gaussian curvature analysis: AAPG Bulletin, v. 78/12, p

3 Delineating Fractures from Seismic Attributes Satinder Chopra Kurt J. Marfurt ARCIS CORPORATION, CALGARY UNIVERSITY OF OKLAHOMA, NORMAN

4 Qualitative interpretation of fractures Fractures enhance permeability in reservoirs and so impact the production and recovery efficiency in those areas. Consequently, detection and characterization of fractures in reservoirs are of great interest. Surface seismic data have been used for detection of faults and large fractures. Recent developments in seismic attributes have shown promise in identifying closely spaced fractures, or interconnected fracture networks. How can we interpret fractures from discontinuity attributes?

5 Coherence High 2 km Low Horizon slice through the coherence volume Data courtesy: Arcis Corporation, Calgary

6 Curvature Positive curvature (Concave downward) r Negative curvature (Concave upward) r Fault displacement Positive curvature r r Negative curvature

7 Curvature 280 ms Segment of a seismic section after structure-oriented filtering Seismic section with most-positive curvature overlaid on it. Neg Pos

8 Curvature 300 ms Seismic section with most-positive curvature (long-wavelength) overlaid on it. Seismic section with most-positive curvature (short-wavelength) overlaid on it. Neg Pos

9 Curvature 300 ms Seismic section with most-negative curvature (long-wavelength) overlaid on it. Seismic section with most-negative curvature (short-wavelength) overlaid on it. Neg Pos

10 Coherence High 2 km Low Horizon slice through the coherence volume Data courtesy: Arcis Corporation, Calgary

11 Curvature Pos 2 km 0 Neg Horizon slice through the most-positive curvature volume Data courtesy: Arcis Corporation, Calgary

12 Curvature Pos 2 km 0 Neg Horizon slice through the most-negative curvature volume Data courtesy: Arcis Corporation, Calgary

13 Coherence and Curvature 2 km High Pos 0 Low Neg Color stack of coherence, most-positive curvature, and most negative curvature Data courtesy: Arcis Corporation, Calgary

14 Seismic Attributes for Fault/Fracture Determination 1. Conditioning of data 2. Choice of algorithm 3. Use of dip-steering option

15 Dip-steered median filtering 1430 ms 1790 ms 1430 ms 1790 ms Inline sections through (a) the input seismic volume, and (b) the median filtered seismic volume 3 km Strat slices through coherence volumes run on (c) the input seismic volume, and (d) the median filtered seismic volume, 76 ms below the horizon shown in (a) and (b)

16 Structure-oriented filtering 1200 ms Neg 0 Pos 2100 ms Input seismic 2 km Input seismic after pc-filtering (Data courtesy: Olympic Seismic, Calgary)

17 Structureoriented filtering vs median filtering (a) Neg 0 Pos (b) Input data (c) 3-point dip-steered median-filter Comparison of seismic time slices at 1778 ms PC-filter 1.5 km

18 Structureoriented filtering vs median filtering Neg Comparison of horizon slices, 66 ms above a flattened marker 0 Pos Input data 3-point dip-steered median-filter PC-filter 1.5 km

19 Common causes of acquisition footprint Problems due to acquisition program Non-uniform offsets and azimuths in bins Non-uniform backscattered noise suppression Obstacles such as lakes, villages, or platforms Currents and tides Problems due to processing Incorrect velocities Migration operator aliasing

20 Footprint removal (a) (b) Low High (c) (d) (Chopra and Larsen, 2000)

22 Coherence algorithm evolution Seismic Crosscorrelation Canyon Salt Semblance Eigenstructure Channels (Gersztenkorn and Marfurt, 1999) Notes by Presenter: Comparison of alternative coherence algorithms used on data from South Marsh Island, Gulf of Mexico, USA. The volume shown contains both structural and stratigraphic features associated with deposition over a terrain influenced by salt tectonism. (a) A time slice through the time-migrated seismic data at 1.8 s. S denotes a salt dome, and F indicates several radial faults. Corresponding slices through coherence cubes were generated using the (b) three-trace crosscorrelation algorithm, (c) a five-trace semblance algorithm, and (d) a five-trace eigenstructure algorithm. All coherence computations used the same 80-ms vertical analysis window. The circular rings seen in (a) correspond to sediments dipping against a salt dome, which are cut by radial faults indicated by F. The disorganized feature indicated by C in the northeast is interpreted to be a canyon. The salt dome and faults appear to be incoherent (black) in (b) through (d). Note that there is considerably less speckle noise in the five-trace semblance algorithm than in the three-trace crosscorrelation algorithm. An even greater improvement in the signal-to-noise ratio and in lateral resolution accompanies the five-trace eigenstructure algorithm. The structural artifacts (leakage) about the salt dome indicated by L and the overall grayer level of the image in (d) are the result of a failure, in this early work, to search over structural dip. After Gersztenkorn and Marfurt (1996).

23 Comparison of algorithm performance Coherence (Eigen decomposition) Coherence (Energy ratio) Time slices at 1342 ms

24 Comparison of algorithm performance Coherence (Eigen-decomposition) Coherence (Energy ratio) Coherence (Eigen-decomposition) Coherence (Energy ratio) Time slices

26 (a) (b) 3 km Coherence using semblance without dip-steering option Coherence using Energy ratio Low High (Data courtesy: Olympic Seismic, Calgary)

27 (a) Low High Semblance (b) Coherence (Energy ratio) after PC-filtering

28 (a) Strat-slices (b) Semblance without dip steering Energy ratio with dip steering Low High

29 (a) Time slices (1360 ms) (b) Semblance without dip steering Energy ratio with dip steering Low High

30 (a) (b) Seismic Time slices (1240 ms) Coherence (semblance) Neg 0 Pos Low High (c) Coherence (Energy ratio) (Data courtesy: OILEXCO, Calgary)

31 Semblance on input data Semblance on input data with dip-steered median filter Energy-ratio on PC-filtered data

32 Seismic Attributes Volumetric computation of curvature Horizon-based curvature has been used successfully for prediction of faults and fractures (Lisle(1994), Hart et al. (2002)). Volumetric curvature dispels the need to pick horizons (Al-Dossary and Marfurt ( 2006).

33 A Neg Pos A 1240 ms 1520 ms (b) A 2.5 km (c) A (d) A (e) A A A A A Neg 0 Pos Neg 0 Pos Neg 0 Pos Neg 0 Pos A segment of a seismic section along a random line (shown dotted in Figure 1(b)) showing a horizon (in yellow), (b) most-positive curvature computed on the horizon, (c ) mostpositive curvature extracted along the horizon from the most-positive curvature attribute volume, (d) most-negative curvature computed on the horizon, (e) most-negative curvature extracted along the horizon from the most-negative curvature attribute volume. Notice: the artifacts seen on the horizon computed curvature displays are not seen on the attributes extracted along the horizon from the curvature attribute volumes.

34 Coherence Most-positive (long-wave) Most-negative (high-res) Time slice 1160 ms

35 Strat-slices Most-positive curvature (Long wavelength) Most-positive curvature (High resolution)

36 Seismic Attributes For delineation of fractures, the frequency content of input seismic data is important

37 Pre-stack time migrated Amplitude data Relative acoustic impedance derived from thin-bed reflectivity 50m thick carbonate reef Platform carbonate 600 m

38 250 ms Seismic (a) (b) (c) Thin-bed reflectivity Reflectivity with a bandpass wavelet (high end 120 Hz) Data courtesy: Arcis Corporation, Calgary

39 High Low Coherence on the input volume Coherence on the frequency-enhanced volume Data courtesy: Arcis Corporation, Calgary

40 Most-negative curvature (short-wavelength) on the input volume Most-negative curvature (short-wavelength) on the input volume with thin-bed reflectivity run on it and then convolved with a wavelet of high-end bandpass of 120 hz

41 Bottomline It is possible to glean meaningful information from seismic attributes. Be aware of how to do it. Fractures can be delineated by running appropriate seismic attributes on seismic data with optimum frequency bandwidth.

42 THANK YOU