You are on page 1of 5

Geological process controlled interpretation based on 3D Wheeler diagram generation

Erik Monsen, Hilde G. Borgos, Pierre Le Guern and Lars Sonneland, Schlumberger Stavanger Research
Summary
The techniques of a new seismic stratigraphic interpretation
paradigm are presented whereby geology is made explicit
and is brought into the interpretation process through
interactive use of an automatically generated 3D Wheeler
diagram. The Wheeler diagram explicitly captures the
geological context throughout the interpretation process,
and provides a great tool for fast QC and data
understanding.
Introduction
Chronostratigraphy is the analysis and subdivision of the
stratigraphic record into bodies of sediment or rock

represented by a particular age, separated from underlying


and overlying units by isochronous surfaces. Wheeler
(1958) introduced the chronostratigraphic charts, in which
stratigraphic cross sections are plotted with a vertical
geological time axis rather than a thickness axis. The
Wheeler diagrams are a useful way of indicating the actual
range in age from place to place of stratigraphic units and
unconformities.
Previous use of Wheeler diagrams has predominantly been
limited to 2D cross sections due to the complexity of the
analysis and time it takes to generate them. However,
recent advances in computer performance and image
processing techniques have paved the way for automated
Wheeler diagram construction, not only in 2D but also in

Figure 1. 3D Wheeler diagram. (Top left) All stratigraphic events are extracted from the seismic data and categorized using their local signal
characteristics. (Top right) The 3D Wheeler diagram visualizes all extracted events using calculated relative geological time as their vertical
coordinate. (Bottom right) The total areal extents of all events are plotted for each relative geological age. (Bottom left) This can be used for
filtering and displaying events as a function of time of deposition.

SEG/San Antonio 2007 Annual Meeting

885

Geological process controlled interpretation

3D (Keskes, 2004; Stark, 2004, 2005; de Groot, 2006). In


this paper we show a new approach to automating and
extending stratigraphic analysis and interpretation
techniques to 3D using horizon patches. New tools for
automated generation and interactive use of 3D Wheeler
diagrams (Figure 1) are developed, and represent cutting
edge research into new interpretation paradigms. Potential
applications go beyond quality control of seismic
interpretation by opening new perspectives on basin scale
analysis for source rock, hydrocarbon migration and trap
identification.
Simultaneous extraction of all stratigraphic events
A Wheeler diagram is built by identifying key stratigraphic
events and assigning relative time of deposition according
to their relative placement in the stratigraphic record.
Direct measurements of this information can be obtained
from well logs, but inherently, this represents a very local
measurement. Seismic data, on the other hand, is an
indirect measurement of lower resolution, but provide us
with observations of the stratification over large areas.
Careful analysis of the 3D seismic data is required in order
to identify and characterize the individual stratigraphic
events. Not only is it important to capture the structural
framework, it is also important to capture waveform

variations along the horizons as this potentially carries


information on sub-seismic depositional events, erosional
events or areas of non-deposition.
Classification techniques can be applied to automatically
group seismic events into classes of similar waveform, and
this methodology is in the following referred to as extrema
classification (Borgos, 2003, 2005). To perform this
classification, first all extrema (minima and maxima) in the
seismic cube are detected. The calculations use a
polynomial reconstruction of each seismic trace, providing
an analytical expression of the trace that enables the
extrema to be detected with sub-sample precision
(Sonneland, 1998). Next, a set of waveform attributes is
calculated at the extrema positions, once more based on a
polynomial reconstruction of the trace. The waveform
attributes have single point support. Still, the set of
attributes at the exact extrema position contains overall
information about the waveform in a vertical window
around the extrema. The extrema and attribute calculations
result in a set of sparse 3D cubes with values only at
extrema positions. A 3D unsupervised or supervised
classification is then run on these cubes. The output is a
sparse 3D cube containing class labels along all seismic
horizons.
The extrema classification allows automatic extraction of

Figure 2. 3D Wheeler diagram construction. All stratigraphic events are extracted from the seismic data and categorized using their local signal
characteristics. Analyzing the over/under relationships among the events along each trace (upper left) using a directed graph (upper right) enables
analysis and calculation of relative geological ages for all events (bottom right) This is used for generating a 3D Wheeler diagram (bottom left),
where all extracted events are visualized using the calculated relative geological time as their vertical coordinate.

SEG/San Antonio 2007 Annual Meeting

886

Geological process controlled interpretation

seismic interfaces based not only on connectivity in 3D, but


also on homogeneity in signal shape. This is achieved by
extracting geometry primitives (referred to as 3D extrema
patches) from the sparse 3D class cube, allowing each
primitive to grow along one class only. A change of class
reflects a change in waveform, thus the class information
provides a constraint for growing only where the signal
shape does not change too much.
Relative dating of stratigraphic events
According to the basic principle of superposition, if one
sedimentary layer is stacked on top of another sedimentary
layer, then it must be younger than the one below. This is
of course an idealized notion, but nevertheless a valid one
in many circumstances.
Following the simultaneous extraction of all stratigraphic
events from the 3D seismic data, it is now possible to
analyze the over/under relationships of the events in order
to determine their relative time of deposition. This is seen
in Figure 2 displaying a workflow for generating a 3D
Wheeler diagram from the previously extracted set of
geometry primitives. In this workflow, the basic structure
used for calculating a stratigraphic events relative
geological age is a directed acyclic graph. By traversing

the individual traces of the volume and recording the


over/under relationships among the extracted horizon
patches, a directed graph is built where the nodes represent
the extrema patches and directed edges are used to indicate
relative age dependencies (i.e., if patch A precedes patch B,
that is, is shallower than patch B, a directed edge is inserted
from node A to node B). Topologically sorting this graph
produces a linear ordering of the patches. Assigning an
initial relative geological age to the first/youngest patch in
this linear ordering, the relative age of all other patches can
be found by visiting the nodes according to this linear order
and assigning incremented relative geological ages at each
step (selecting the maximum in case of multiple
assignments).
The 3D Wheeler diagram can now be generated by
flattening all stratigraphic events (i.e., their 3D extrema
patches) and plotting them in 3D with the calculated
relative geological time as their vertical value. This renders
an interactive 3D Wheeler diagram, as illustrated in the
lower left part of Figure 2. Figures 1 and 3 show examples
of how the 3D Wheeler diagrams would look for real data.

Figure 3. Directed acyclic graphs are used to calculate relative geological ages. This figure shows an example of how such a graph (right) might
look. Following topological analysis and propagation of relative geological ages, the (upper left) data set is transformed into a 3D Wheeler
diagram (lower left) by re-rendering the extrema patches with the calculated relative geological age as their new vertical coordinate.

SEG/San Antonio 2007 Annual Meeting

887

Geological process controlled interpretation

Dual-domain interpretation system


Visualizing the stratigraphic events in both the seismic
domain and the Wheeler domain enables efficient dualdomain interactions, such as cursor tracking, event
selection and filtering. New interaction modes and
information displays can now be developed. One such
example is the display seen in the lower right hand side of
Figure 1 where the total areal extent of all sediments are
plotted for the different geological ages, thereby revealing
important events, such as sequence boundaries. The
stratigraphic events can now be filtered and sorted
according to this information.
Conclusions
This work demonstrates the enabling techniques of a new
interpretation paradigm where geology and geological
processes are made explicit through automated generation
and interactive use of 3D Wheeler diagrams.
Classification of seismic extrema is applied to generate
geometry primitives for structural interpretation. The
seismic volume of interest is then represented through a set
of extrema patches. These primitives are signal consistent,
extracted with sub-sample precision, and contain attribute
information along with the geometry. The interpreter works
interactively with these primitives to build one or multiple
horizons (Borgos, 2005).
A 3D Wheeler diagram is built by capturing all extrema
patches and their over/under relationships in a directed
acyclic graph, the subsequent topological analysis of which
reveals the relative geological ages (time of deposition) of
stratigraphic events. The Wheeler diagram can is kept
current by seamlessly and in real-time updating the
calculation of relative geological ages in response to user
interactions and interpretations.
The capture and explicit representation of geological
processes in a Wheeler diagram provides an excellent
means of understanding sediment distribution and sorting
processes, which in turn can be utilized for better property
population of geological models.

Acknowledgements
We thank Petrobras/PRAVAP IOR program for permission
to publish this data set.

SEG/San Antonio 2007 Annual Meeting

888

EDITED REFERENCES
Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2007
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.
REFERENCES
Borgos, H. G., T. Skov, T. Randen, and L. Sonneland, 2003, Automated geometry extraction from 3D seismic data: 73rd Annual
International Meeting, SEG, Expanded Abstracts, 15411544.
Borgos, H. G., T. Skov, and L. Sonneland, 2005, Automated structural interpretation through classification of seismic horizons,
in A. Iske and T. Randen, eds., Mathematical methods and modelling in hydrocarbon exploration and production:
Springer Verlag, 89106.
de Groot, P., G. de Bruin, and N. Hemstra, 2006, How to create and use 3D Wheeler transformed seismic volumes: 76th Annual
International Meeting, SEG, Expanded Abstracts, 10381042.
Keskes, N., S. Guillon, M. Donias, P. Baylou, and F. Pauget, 2004, Method of chrono-stratigraphic interpretation of a seismic
cross section or block: U. S. Patent US6771800.
Sonneland, L., P-O. Tennebo, T. Gehrmann, O. Yrke, K. S. Boge, and G. Berge, 1998, A method of processing seismic data
signals: PCT Patent WO199837437.
Stark, T. J., 2004, Relative geologic time (age) volumes - Relating every seismic sample to a geologically reasonable horizon:
The Leading Edge, 23, 928932.
, 2005, Generation of a 3D seismic "Wheeler Diagram" from a high resolution age volume: 75th Annual International
Meeting, SEG, Expanded Abstracts, 782785.
Wheeler, H. F., 1958, Time-stratigraphy: Bulletin, 42, 10471063.

SEG/San Antonio 2007 Annual Meeting

889

You might also like