You’ve heard it before:
Three-D seismic is a blessing there’s so much data, and it’s a curse
there’s so much data.
This may be a truism, however, that has outlived
A whole new technology has emerged that can tear
into massive volumes of 3-D data to not only develop a play concept,
generate leads and qualify prospects, but to do so at something
akin to lightening speed.
It’s been used, for instance, to identify a potentially
productive stratigraphic feature from a volume of seismic data encompassing
50 offshore blocks, scanning the whole shebang in 32 hours to highlight
and automatically map all the look-alikes in the data volume for
that particular reservoir type.
The technology is called ImageGenetics™, a pattern
recognition process that traces its origin to the biological sciences.
and here’s the blueprint:
The human genome is made up of only four letters
of the alphabet, which combine with one another to form a pattern
that then combines with other patterns to form patterns of patterns,
or textures. It’s this hierarchy of increasing complexity that lets
the four simple features build the entire human genome.
Recognizing the potential to unravel other complex
data sets by looking for the same type of pattern repetition using
the same sort of hierarchal structure of DNA, Fred Young at Chroma
Graphics invented and patented the ImageGenetics process.
Along with the computer graphics and medical fields,
the oil and gas industry was quickly identified as an ideal market
for the technology.
This encouraged the company to organize its Chroma
Energy subsidiary, which comprises a team of oil finders focused
on mining large volumes of data to generate qualified drilling prospects.
The technology is different from the standard approach
the geophysicist has typically taken, said Peter Duncan, chief operating
officer at Chroma Energy.
“We’ve always broken the seismic data down into attributes
such as amplitude, instantaneous phase or whatever,” Duncan said,
“and draped the attribute over the structure to see what the distribution
of that single attribute tells us about the geology.
“What Chroma does is define a new set of attributes
that are data derived and made up of snippets of the data itself
- just like the biological snippets used to analyze DNA - and look
for patterns of those attributes and then textures of those patterns,”
he said. “And that’s the significant leap forward.”
Dallas-based consulting geophysicist Mike Forrest
said the pattern recognition technology “enhances the value of 3-D
data sets with detail interpretations of geomorphology and stratigraphy
that are not possible using standard time slices.”
The initial step in implementing the technology is
to use an analog or play concept and derive a set of patterns and
textures that highlight and illustrate how a play concept exhibits
itself within the data.
“I can extract unique patterns and find a type code
that lets me see my perception of a play concept,” said Peter Whitehead,
president and CEO of Chroma Energy. “The images we extract of the
geology are more definitive than ever before.
“We visualize the textures and patterns with a full
32-bit voxel visualizer and false color imagery to reassemble multi-attribute
combinations of these textures and patterns in a powerful, illuminating
way,” he continued.
“It’s like giving the interpreter a paint brush to
paint the geology onto the textures and patterns that now can be
recognized in the seismic data.”
Once the geology of interest is identified, the computer
rapidly locates similar occurrences in the data set to quickly create
an unqualified lead list. The search is then further refined by
drilling down into the data, telling the computer, for example,
to isolate the sand bodies and divide each thickness into different
colors to show structure within a sand body itself.
By taking something that looks the same and breaking
it apart into constituent pieces, the interpreter can begin risking
such variables as the kind of communication likely to be encountered
within the reservoir, and using it to qualify leads and prospects.
The technology is garnering some true believers in
the real world of E&P.
“What I like about it is it takes me where I haven’t
been before in looking at the pattern data, relationships it finds
and how it combines different patterns of characters to identify
things in seismic not in a normal view,” said Michael Hiner, chief
geophysicist at Unocal’s Spirit
“You can interrogate a pattern data set to look at
just thin beds, for instance, and with just a few motions on the
menu it will quickly highlight those events for the whole volume.
It’s very powerful.”
So far, Hiner has used the tool on three Gulf of
“We’re still evaluating two, and on the other one
we’ve come to some conclusions and found the tool did as we tasked
it to give us solutions we didn’t have in a classical interpretation,”
he said. “What remains is to test those conclusions against the
ImageGenetics™ might be used along with some of the
other seismic packages on the market that are employed to highlight
stratigraphy, such as Coherence technology.
“Coherence looks for discontinuities and highlights
the edges of bodies, and you infer what’s in between those edges,”
Duncan said. “Our tool looks for similarities and paints the geobodies,
so it’s exactly the complement to Coherence.”
It’s also a complement to the tried-and-tested Stratimagic
application, according to Hiner, who used both techniques on the
three GOM Unocal properties.
Still, there are some fundamental differences in
the two technologies:
- Stratimagic is a mapping tool that requires the user to first
identify a horizon or zone of interest to be evaluated. The application
is based on Neural Networks, which is a branch of artificial intelligence
computing that addresses mainly shape recognition.
- ImageGenetics™ is a volume tool that looks at a whole volume
of seismic data with no preparatory work. The interpreter drives
the process, rather than a Neural Net.
“Our goal and intent,” Hiner said, “was to learn
what the tools can do, test them against each other and see if we
get similar results and if there’s a breakout in stratigraphy in
one over the other.”
Hiner noted he would use both tools on the same project
today because of the evolutionary state of the new technology.
“Stratimagic is established, vetted by our company
and others in the industry, and it’s a reliable tool that everybody
understands,” Hiner noted. “Still, if I were walking into a new
survey for the first time where I didn’t have a specific horizon
I was interested in, I would use the Chroma tool to look at the
whole volume without having to do anything more than set it up.
“We learned that both Stratimagic and the Chroma
tool are able to identify very subtle stratigraphy and enhance our
view of that stratigraphy,” he continued, “and give us internal
details on some of the stratigraphic features that we could not
get with other more classical interpretations.”
To experiment, they used some channel geometries
they had picked using StratAmp in Landmark that were “quite exciting,”
but the internal detail around the channels couldn’t be seen.
When Stratimagic was used on one of the intervals,
it was able to look inside the channel and show changes in wavelet
character within the channel geometry and also outside, Hiner said.
The Chroma tool was next used, and “we ran it blind,
not telling the Chroma staff what we were looking for,” he continued.
The geoscience staff at Chroma detected the channels,
focused in on them and then, “in a horizon sense generated the same
kind of detail that we had gathered in Stratimagic,” Hiner said.
In a break with the traditional approach taken by
service companies to sell or license seismic application software,
Chroma intends to keep its new tool close to the vest.
“We’ll keep the ImageGenetics algorithm and ChromaVision
tool proprietary to our clients who partner with us on projects
where we share in the risk and proportionate reward,” Whitehead