See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268454885

Structural mapping with spectral attributes

Conference Paper · September 2012

DOI: 10.1190/segam2012-0435.1

CITATIONS READS

12 199

1 author:

Andreas Laake
Schlumberger Limited
107 PUBLICATIONS 451 CITATIONS

SEE PROFILE

All content following this page was uploaded by Andreas Laake on 30 September 2015.

The user has requested enhancement of the downloaded file.

 Structural mapping with spectral attributes
Downloaded 11/28/12 to 136.249.13.11. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Andreas Laake*, WesternGeco

Summary Methodology

Information may be extracted from seismic data cubes
using attributes that represent specific characteristics of the
data contained in the cube. Examples of typical attributes
are dominant frequency, RMS amplitude, and spatial
variance. The description of complex geological features
usually requires information from multiple attributes. One
possibility for simultaneous attribute analysis is attribute
triplets in continuous-color representation as red, green,
blue (RGB) or hue saturation value (HSV). We describe a
method to sharpen the continuous-color representation of
multiple attributes with the goal of providing structural
lineaments for geological interpretation. We represent three
different attributes in RGB and extract the structural, which
is contained in the boundaries of the colors in the RGB
image.
The primary results are cubes containing structural
lineaments on the one hand and the structure-sharpened
sharpened RGB images on the other hand. The technique
may also be applied to combinations of more than three
attributes.
The technique concerns a method to sharpen the
continuous-color representation of multiple attributes with
the goal of providing structural lineaments for geological
interpretation. The seismic data required to apply this
technique are configured in a cube. In the first step,
multiple attributes are extracted from the seismic cube
using either commercially available or specific software.
Such attributes may comprise signal properties such as
amplitude and frequency, structural properties such as
curvature and gradient, or statistical properties such as
variance. Each of these attributes is enhanced to optimize
the further analysis.
Next, three attributes are selected and allocated to the red,
green, and blue bands in a continuous-color RGB
representation. The dynamic range of the data is, thereby,
extended to the third power of the resolution of each single
attribute. This increase in dynamic range allows image
processing to be applied to the data.

Introduction

Extraction of attributes from seismic data is provided

routinely by existing geological modeling software.
Frequency attributes are among the most often-used
characteristics. Spectral decomposition and geological
interpretation of spectral data were described by Partyka et
al. (1999) and Dunbar et al. (2000). The representation of
attribute triplets in RGB and its geological interpretation
was described by Bucker (1989), Bucker et al. (1990), Dorn
(2002), Chopra and Marfurt (2007, 2008a, 2008b), Chopra
et al. (2010), and Chopra et al. (2011).

For structural interpretation of seismic data cubes, several

attributes are extracted and usually compared in separate
displays. This approach makes it difficult to extract
structural information from multiple attributes
simultaneously. Our goal was, therefore, to develop a
method that merges multiple attributes and allows the
extraction of structural lineaments from the simultaneous
analysis of multiple attributes in continuous-color
representation. We will describe the methodology and
apply the technique to structural delineation within
individual layers as well as to structure correlation between
several layers. We chose an erosional structure from
Australia to demonstrate the technique.
Figure 1: Generic workflow for structure-sharpened RGB analysis
of multiple seismic attributes.
The RGB image codes physical properties such as
frequency, amplitude, or velocity as color shades.
Structural information is contained in the boundaries
between areas of different color and amplitude. The
conversion from RGB to HSV separates the pure color
information in the hue band from the color-independent
intensity information in the saturation band. The saturation

© 2012 SEG DOI http://dx.doi.org/10.1190/segam2012-0435.1

SEG Las Vegas 2012 Annual Meeting Page 1
 Structure from spectral attributes
band gradient provides edges, which represent structural
Downloaded 11/28/12 to 136.249.13.11. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
lineaments contained in the data. In the final step, the
structural lineaments are convolved with the previously
generated RGB image, thus providing a structurally
sharpened RGB image of this attribute combination.
Usually, such an image represents the geological character
of the area of investigation more realistically than the
description by one single attribute alone. Figure 1 shows
the generic workflow for structurally sharpened
multiattribute analysis in continuous color RGB.
We will now illustrate each of these processing steps on
time-slice seismic data from an erosional structure in
Australia (Figure 2). We begin by band-pass filtering the
seismic data. We generate nine bands of 10 Hz bandwidth
each and select one high frequency (60 Hz), one middle
frequency (30 Hz), and one low frequency (10 Hz) band
each (Figures 2a, 2b, and 2c). The three frequency slices
are considered the input attributes.
Figure 2: Illustration of individual steps in the workflow: band-
pass filtered seismic time slice for 60 Hz (a), 30 Hz (b), and 10 Hz
(c); (d) representation in RGB; (e) RGB converted to HSV; (f)
saturation band of HSV image. The results are: (g) edge detection
of the saturation band; (f) as grayscale contours; (h) the
convolution of the edge; (g) with the RGB image (d) in a
structurally sharpened RGB image.
We allocate the 60-Hz slice to red, the 30-Hz band to
green, and the 10-Hz band to blue in a continuous-color
RGB image (Figure 2d). This image combines the
characteristics of the three input attributes in one RGB
image. Because we aim at extracting the structural
information from the attributes, we convert the RGB image
to HSV (Figure 2e) because the HSV image represents the
information contained in the color in the hue band;
© 2012 SEG
SEG Las Vegas 2012 Annual Meeting
whereas, the color-independent intensity information is
represented in the saturation band. We extract the
saturation band, run edge detection on the intensity data,
and display the edge contours in grayscale (Figure 2g).
These contours represent structural lineaments contained in
this single time slice. We will later see that the elongated
structure represents the subcrop boundary around an eroded
anticlinal structure. In the final step, we convolve the edge
(Figure 2g) with the RGB image (Figure 2d). The result is a
structure-sharpened RGB image that reveals the spectral
structure of the seismic data within the time slice (Figure
2h).
For validation, we extract the time slice and corresponding
section from the seismic cube. We plot them on top of each
other such that the bottom of the time slice fits the top of
the section (Figure 3). The section shows an anticlinal
structure, the top of which is eroded. On both flanks of the
anticline, flat continuous layers dip away from the
anticline.
Figure 3: Correlation of time slice and section in conventional
amplitude (a) and RGB from 60, 30, and 10 Hz (b).
Conventional data are displayed as amplitude in grayscale
(Figure 3a) and are compared with the RGB image created
from the 60-, 30-, and 10-Hz frequency bands. The RGB
images more details in both the time slice and the section.
For the last implementation, we choose the spatial variance
attribute, which highlights areas of significant lateral
change in the seismic cube and is well-suited to reveal
sudden vertical changes. We extract the spatial variance
attributes for frequency bands of 60, 30, and 10 Hz, which
we used in the RGB analysis above, and generate the
structure-sharpened variance RGB image. Figure 4 shows
DOI http://dx.doi.org/10.1190/segam2012-0435.1
Page 2
 Structure from spectral attributes

selected steps in the workflow generating the structurally

Downloaded 11/28/12 to 136.249.13.11. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

sharpened RGB image for the spatial variance.

Figure 4: Generation of structure-sharpened RGB for spatial

variance attribute obtained from 60-, 30-, and 10-Hz frequency
bands. 30-Hz variance time slice (a), 60-, 30-, and 10-Hz variance
RGB (b), edge of HSV saturation band (c), and structurally
sharpened spectral variance RGB (d).

Results and discussion

For validation, we extract three seismic horizons from the

seismic cube (Figure 5): one above the erosional
discontinuity horizon (blue), one on this horizon (green), b
and one below (red). We will drape the results from the
structural attribute analysis onto these horizons.

Figure 5: Location of horizons in the data cube and main structural
orientations of the anticline. The green arrows indicate the
direction to north.
c
Figure 6: Result from structurally sharpened attribute RGB images
draped on a set of three horizons around the erosion horizon and
correlation with conventional seismic inline and crossline sections.
Conventional RMS amplitude in grayscale coloring (a),
structurally sharpened spectral amplitude RGB (b), and variance
attribute RGB (c). The latter are computed for 60-, 30-, and 10-Hz
frequency bands. The main axes of the anticline – NW-SE and
SW-NE – are shown as arrows below the erosion horizon. On
either flank of the anticlinal structure, we observe flat layers
pinching out towards the top of the anticline. The green arrows
indicate the direction to north.

© 2012 SEG DOI http://dx.doi.org/10.1190/segam2012-0435.1

SEG Las Vegas 2012 Annual Meeting Page 3
 Structure from spectral attributes
The results of structurally sharpened RGB attribute analysis
Downloaded 11/28/12 to 136.249.13.11. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
are draped over the horizons around the erosion horizon of
the anticline (Figure 6). We have partially cut away the
horizons from front to back to reveal the structural changes
from layer to layer. The horizon below the erosion extends
across the entire cube, the erosion horizon itself is cut away
by about 50%, and the remaining analysis surface is
covered by the horizon above the erosion horizon.
Figure 6a shows the grayscale coding of the instantaneous
amplitude; this is the usual way of coding and displaying
seismic data. For orientation, four crossline sections were
inserted at the lines where the horizons were cut. The
amplitude display reveals only the outline of the anticline
(yellow dashed line), but does not show details.
Figure 7: Comparison of structural characterization from
instantaneous amplitude (a), and the merge of spectral RGB
amplitude and variance (b).
The spectral RGB from 60, 30, and 10 Hz (Figure 6b)
clearly reveals the subcrop around the anticline, which
© 2012 SEG
SEG Las Vegas 2012 Annual Meeting
results from the pinch-out of the layers on the flanks of the
anticline. It is highlighted by the strong black signature.
Within the anticline, structures are revealed in the layers
above and on the erosion horizon; whereas, the horizon
below the erosion does not reveal any noticeable structures.
The spectral variance RGB showing the spatial distribution
of variations in the 60-, 30-, and 10-Hz frequency bands
maps mainly small-scale structures and lineaments (Figure
6c). These small structures are mainly observed on the
erosion horizon and below, and may correlate with fluvial
channels. Sets of lineaments sub-parallel to the main
anticlinal axis are observed across the SE part of the
anticline. They may represent local fault zones. The layer
above the erosion horizon contains larger-scale structures
that are identical to the structures mapped by the spectral
RGB.
The use of spectral attributes – whether as amplitudes or as
spatial variance attributes – appears beneficial for structural
mapping of geological structures. The instantaneous
amplitude, which consists of the spectral sum of the
amplitudes of the individual frequencies, does not reveal
subtle structures because of interference by the individual
frequency bands. The merge of different spectral attributes
in RGB provides the continuous-color coding of the
spectral attribute values in a large dynamic range. Subtle
frequency changes are mapped as color changes and can be
used to distinguish small-scale structural and sedimentary
features. Figure 7 demonstrates how much structural detail
can be gained from using spectral attributes (Figure 7b)
compared to the standard approach using the instantaneous
amplitude (Figure 7a).
Conclusions
The extraction of spectral attributes from seismic data and
representation in continuous colors reveals structures that
cannot usually be mapped by conventional instantaneous
amplitude processing. The extraction of structural
lineaments from the HSV attribute cube further enhances
the potential for structural delineation by seismic data.
Spectral RGB analysis of amplitude data generally reveals
larger structures, the internal structure of which can be
mapped using spectral-variance RGB analysis. The
combination of these complementary data sets provides a
detailed structural insight for all scales.
Acknowledgements
The author thanks Phil Bilsby and Kevin O’Sullivan for
processing the seismic PSTM cube, Peter Wang for the
discussion of the attributes, and WesternGeco for
permission to publish the data.
DOI http://dx.doi.org/10.1190/segam2012-0435.1
Page 4
 http://dx.doi.org/10.1190/segam2012-0435.1

EDITED REFERENCES
Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2012
SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for
each paper will achieve a high degree of linking to cited sources that appear on the Web.
Downloaded 11/28/12 to 136.249.13.11. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

REFERENCES
Bucker, R. H., 1989, Method for continuous color mapping of seismic data: U. S. Patent 4,843,599.
Bucker, R. H., K. J. Marfurt, and T. D. Stanley, 1990, Method for color mapping geophysical data: U. S.
Patent 4,970,699.
Chopra, S., and K. J. Marfurt, 2007, Seismic attributes for prospect identification and reservoir
characterization: SEG Geophysical Developments Series No. 11.
Chopra, S., and K. J. Marfurt, 2008a, Seismic attributes for stratigraphic feature characterization: 78th
Annual International Meeting, SEG, Expanded Abstracts, 1590–1594.
Chopra, S., and K. J. Marfurt, 2008b, Emerging and future trends in seismic attributes: The Leading
Edge, 27, 298–318.
Chopra, S., K. J. Marfurt, and S. Misra, 2010, Seismic attributes in frequency-enhanced seismic data:
80th Annual International Meeting, SEG, Expanded Abstracts, 1462–1466.
Chopra, S., S. Misra, and K. J. Marfurt, 2011, Coherence and curvature attributes on preconditioned
seismic data: The Leading Edge, 4, 386–393.
Dorn, G. A, 2002, Computing and visualization: The Leading Edge, 6, 581–586.
Dunbar, J. A., P. A. Allen, and P. D. Higley, 2000, Color-encoding multifrequency acoustic data for near-
bottom studies: Geophysics, 65, 994–1002.
Partyka, G., J. Gridley, and J. Lopez, 1999, Interpretational applications of spectral decomposition in
reservoir characterization: The Leading Edge, 3, 353–360.

© 2012 SEG DOI http://dx.doi.org/10.1190/segam2012-0435.1

SEG Las Vegas 2012 Annual Meeting Page 5
View publication stats