Are geologists drowning in data?
How good are our analytical models?
Even when we know what we're doing, do we know that we're doing the right things?
AAPG author, course teacher and lecturer Norman S. Neidell thinks he has the answers to those questions, which are "Probably," "Not very" and "Too many times, we don't."
Neidell, a well-known seismic expert and geophysical consultant, will present his views in the paper "Geologic Information and Seismic Data -- How Much Do We Need?" at the AAPG Gulf Coast Association of Geological Societies meeting in November.
He authored the paper with Rosemary Mullin, an associate in N.S. Neidell & Associates in Houston. Their concern centers on the use of geologic information with seismic data.
When velocity models guide depth imaging in preliminary studies, geologic information is usually sparse or nonexistent.
Approaches that require significant geologic input are not likely to work well, Neidell noted, because of the paucity of inputs.
Also, an abundance of data and detail won't produce good results unless the industry can identify what information matters most -- and then can apply that knowledge effectively, he warned.
"Sometimes we put 15 decimal places on an estimate," he said. "We have to be more humble in saying we don't have good analytical models."
More is not always better, and a sea of data often will overwhelm simplistic models, according to Neidell. And he's concerned that the industry too often looks for quantity of data, instead of quality and understanding.
"I don't necessarily have answers," he said. "But I do have some oversights to point out. I'm trying to bring the industry to a screeching halt and say, ‘What do we really need -- and, how do we get it?'"
Neidell said he began pondering the problem of effective information use after a trip to the Middle East. That journey resulted from a case of plentiful data and deficient results.
"A few years back I was invited to a conference put on in the Middle East," Neidell explained. "The conference was called Super-K, for superconductivity."
A waterflood project had been planned to enhance production. The program looked reasonably simple and the company certainly didn't lack information.
"The reservoirs in this case were some of the most studied reservoirs in the world," Neidell said, "and they had the best seismic money can buy."
Despite an extensive analysis, however, the injection program never really went as planned. The data worked well with the models, but the geology didn't cooperate.
"Their concept was that they'd drill injection wells and inject water into well A and well B, and then more oil would come out of well C," Neidell said. "What happened was that more oil would come out of well F, or all the water would come out but the oil flow would be unaffected."
In examining the effects of injection on well A and well F, sometimes "it looked as though there was a direct pipeline between the two wells," he recalled.
Neidell said the group of experts at the meeting split into two camps. Half immediately had a perfect explanation for the problem -- although each explanation was different, he said. The other half admitted to being baffled.
"I came away from there thinking, gosh, if we make all the measurements we can make, we still might not make the right measurements," he recalled, "and we might not be able to come up with the right answers."
Neidell realized that success might be attributed to good information, but failure rarely was blamed on poor information choices or weak analysis. For the most part, failures were forgotten and nothing was learned.
"If it works, it goes into a paper. If it doesn't work, it goes into a file," he said. "We seem to be plowing headlong on an approach without a lot of thinking. I want a lot more consideration of why we're failing."
In geophysics, he found that boundaries could not be sharply defined using different data sets for exactly the same area. Or, they could not be defined at all.
"The P-wave section and the shear-wave section over what's supposedly the same geology show completely different characteristics. You get to the question: ‘Which one are you going to believe?'" he said.
"No one has ever correlated fully a shear-wave section to a P-wave section, as far as I know," Neidell noted. And he's seen that even small changes in the same data sets can produce widely different interpretations.
As an example, Neidell cited a comparison of two seismic data sets acquired in the same location, but with the frequency band shifted by 25 hertz. Because of that very small change, the same reflectors can't be seen in both sets, he said.
"The fact is that in many of these comparisons you can't even find the same boundaries," he observed, "Even if we put in gradational boundaries, how do we grade them?"
"We don't really have a technology for making fuzzy boundaries," he said. "You get to a point where you ask, ‘How much can we know?' and ‘How much good will it do us?'"
Instead of trying to provide answers, Neidell is more concerned with raising questions. Geophysicists shouldn't purport to have all the answers -- or even most of the answers -- when good information is missing, he believes.
"The true answer is, unless you stick a well in it, you don't really know," he said. "And then all you know is where the well is."
More. Different. Lots.
Today, service companies and seismic contractors conduct much if not most of the research in exploration tools.
"The industry has let the contracting world drive the progress. And the contracting world's motivation is quite different from the rest of the industry," he said.
"Their goal in the past was to get you to use a lot of data and a lot of different data. And, truly, they did want to give you information that was helpful."
Neidell sees no bad guys in this scramble to produce and sell data. He does think the oil industry has "self-destructed in expertise," leaving it without the in-house knowledge to guide and assess contractor work.
Some contractors now take a piece of exploration projects, giving them a better understanding of industry needs, Neidell noted. And some contractors have built up their own reservoir modeling teams.
"Still," he said, "the mindset of the contractors is ‘More is better. Different is better. Lots is better.'"
Nor does Neidell necessarily believe in the eminence of multi-discipline exploration teams. He said companies tended to make teams by grabbing whatever geologists, geophysicists and engineers were available.
"They put them in the same room and told them to work together. The only thing they forgot to tell them is how," he observed. "You got into a situation where you had a lot of fellowship but not a lot of synergy."
Neidell noted that the industry has good simulation tools that combine different types of data. He also said some simple models are effective, although they work best in the simplest geology of starved basins.
"Our problem comes down to building reservoir models that truly reflect reservoir performance," he said. "Maybe the models we're building are too simplistic.
"The big answer is that we need to spend time building the analytic models to put these things together. Simplistic models of geology are not going to solve the problem of how to use the data, and how to use the new types of data we can get now."