Machine Learning Changes the Job Requirements for Geoscientists

When new technology enters the oil and gas scene, talk of layoffs can creep into water-cooler conversations. Will better software and computers replace people, or will they push the industry forward, creating the need for additional staff?

These questions are especially pertinent for geophysicists today, as artificial intelligence and machine learning technologies are processing and interpreting seismic data at record speeds, often delivering results that rival, if not surpass, that of humans. With some software companies calling their platforms a “seismic revolution” by offering real-time data interpretation, geophysicists might question how they will fit into this new, seemingly supersonic world.

“It is now possible for one person to do the job of many,” said Peter Duncan, AAPG Member and president and CEO of MicroSeismic, Inc. “It now takes someone 20 minutes to do a job that used to take two months. So, what happens to the other people who might have done that same job? There could be displacement in the short term.”

Long-term projections, on the other hand, indicate a world of possibilities.

Express Interpretation

Not only has new technology increased the speed of seismic data acquisition and processing, it is making great headway with interpreting data to quickly identify potential sweet spots for oil and gas. Countless interpretation processes for geological features – such as fault and fracture delineation, stratigraphic facies distribution, lithofacies classification and salt delineation are now being evaluated with machine learning, said Rocky Roden, AAPG Member and owner of Rocky Ridge Resources, Inc., an oil and gas consulting firm in Houston. Roden said newer technology utilizing machine learning greatly expedites the interpretation process and, in many cases, enables the evaluation of numerous types of data at once, well beyond the human brain’s limit of three to four types of data.

When interpreting seismic data, geophysicists must often make educated guesses and assumptions about the subsurface, which inevitably leads to gaps in information, Roden said.

Image Caption

3-D visualization of a depositional facies classification using deep learning convolutional neural networks, Taranaki Basin, offshore New Zealand. Additional info: Seismic facies classification using CNN deep learning enables the identification of structural and stratigraphic facies patterns based on deep learning technology. Seismic facies and other patterns present on seismic data, such as potential direct hydrocarbons indicators, multiples, etc., can be identified in a seismic volume given the appropriate training. The 3-D extent of these features can provide significant and valuable information to the interpretation process. Using such methods, the interpreter is able to extract more information faster from the seismic response by allowing the deep learning process to identify similar patterns in the data as the training data provided by the interpreter.

Please log in to read the full article

When new technology enters the oil and gas scene, talk of layoffs can creep into water-cooler conversations. Will better software and computers replace people, or will they push the industry forward, creating the need for additional staff?

These questions are especially pertinent for geophysicists today, as artificial intelligence and machine learning technologies are processing and interpreting seismic data at record speeds, often delivering results that rival, if not surpass, that of humans. With some software companies calling their platforms a “seismic revolution” by offering real-time data interpretation, geophysicists might question how they will fit into this new, seemingly supersonic world.

“It is now possible for one person to do the job of many,” said Peter Duncan, AAPG Member and president and CEO of MicroSeismic, Inc. “It now takes someone 20 minutes to do a job that used to take two months. So, what happens to the other people who might have done that same job? There could be displacement in the short term.”

Long-term projections, on the other hand, indicate a world of possibilities.

Express Interpretation

Not only has new technology increased the speed of seismic data acquisition and processing, it is making great headway with interpreting data to quickly identify potential sweet spots for oil and gas. Countless interpretation processes for geological features – such as fault and fracture delineation, stratigraphic facies distribution, lithofacies classification and salt delineation are now being evaluated with machine learning, said Rocky Roden, AAPG Member and owner of Rocky Ridge Resources, Inc., an oil and gas consulting firm in Houston. Roden said newer technology utilizing machine learning greatly expedites the interpretation process and, in many cases, enables the evaluation of numerous types of data at once, well beyond the human brain’s limit of three to four types of data.

When interpreting seismic data, geophysicists must often make educated guesses and assumptions about the subsurface, which inevitably leads to gaps in information, Roden said.

“We make a lot of assumptions about the data, for example in seismic processing, to get the interpretation in the first place. We assume homogeneity in reservoirs, but in reality, they usually are very heterogenous,” he explained. “Machine learning technology can potentially discover non-linearities in the data that we don’t know in theory. It can improve efficiency and accuracy over time.”

Technology also is eliminating the mundane tasks of the geophysicist, said Deborah Sacrey, AAPG Member and owner of Auburn Energy. “Certain types of machine learning are automating the dull, boring aspects of seismic interpretation, like fault picking, which is time-consuming and laborious,” she said. “So, if people can have that process automated, they can get through large, complex datasets faster.”

The Shifting Role of the Geoscientist

Looking ahead, the role of the geophysicist is beginning to shift in response to newer technology, Sacrey said.

She believes that for upcoming geophysicists to thrive, they will need to adopt specific skillsets.

“They need to be trained in machine-learning technology and data science and statistics. If that’s not on their resumes in the next five years, they may not find a job. It’s an absolute requirement for people coming out of school,” she said. “There is a huge amount of data out there that companies are trying to get their arms around and organize to make it available to geophysicists and geologists in a more efficient manner. Aspects of data science itself is a new opportunity for careers, and I see geophysics moving into that end. There is also a new cadre of jobs being created in the geosciences to help implement machine learning technology in the industry.”

Linda Ford, managing director of programs at the Society of Exploration Geophysicists, believes it is too early to predict how new technology will affect geophysicists in the long run. Yet, she knows one thing for sure: Geophysicists, by profession, are adaptable.

“Geophysicists have been all along data scientists uniquely equipped to be part of digital transformation,” she said. “Machine learning and artificial intelligence will change the nature of the work they do … but it might potentially solve one of the geophysicist’s biggest problems in terms of the common value proposition of their field, which is processing time. This could shift the focus of the geophysicist to more complex problem-solving that requires more critical geophysical thinking not yet adaptable to machine learning and artificial intelligence.”

Geophysicists have already proven their flexibility during the rise of unconventional oil and gas, which required the adoption of reservoir management skills, Ford said. Furthermore, with forward-thinking projects such as carbon sequestration, a geophysicist’s unique perspective on the subsurface could potentially be useful in reducing drilling and environmental impacts, she added.

Ford is even encouraging aspiring geoscientists to look beyond the oil and gas sector.

“We need to help students understand the wide variety of options for careers,” she said. “Geophysics has been so tied to oil and gas. We have a lot of work to do to better spotlight other opportunities with longevity. We are at a critical time with the current emphasis on renewables. There is not an unlimited time for the geosciences in the oil and gas industry.”

New Tech, New Challenges

Having been in the oil and gas business for 44 years and having worked for or consulted with dozens of companies, Roden knows the industry has always been slow to adopt new technology, especially during a downturn.

“I’ve seen too many new technologies come into the industry, and I always hear people say the same thing: ‘People will be losing jobs.’ But it’s usually the opposite. It produces more jobs because the new technology enabled us to do more than we did before,” he said. “Up until now, we really have not seen any machine learning of any significance in the evaluation process in exploration plays. However, most companies do have some sort of initiative to see how machine learning will be a component in their process for determining where to drill wells.”

As operators cautiously enter the field of machine learning and artificial intelligence, Roden said that, rather than being concerned with an immediate displacement of geophysicists, the focus should be on attrition.

“Geoscientists in the industry, they are absolutely shrinking,” he said, attributing the dwindling numbers to retiring Baby Boomers and a lack of hiring during the recent downturn.

Seeing the world’s population triple to 7.7 billion during his lifetime, Roden said the need for energy is ever-growing, and so is the need for geophysicists.

In the eyes of Duncan, ever-evolving technology is what makes geophysics and other fields so attractive. Comparing days when seismic lines were shot a mile apart versus today’s 60 meters, he believes more geophysicists will be needed to handle growing volumes of data.

“At one time, an interpreter might have spent his entire career working in a single area, so he could intuitively interpret the data,” he said. “Now, the computer captures the data, learns from it and does it much more quickly. A geophysicist can move between companies and basins and challenges, and life becomes more interesting.”

As geophysicists get up to speed with machine learning and artificial intelligence, Duncan said that long-term results will be an incredible boost to the industry: “We will be able to dig more deeply into the data and hopefully come up with fewer dry holes, which may well put the drillers out of work!”

Joking aside, Duncan sees a bright future ahead. “We embrace the technology, and the technology embraces us – if you like a challenge.”

vafzsdzfaycqdfzazsetyxda

You may also be interested in ...