Data science and analytics is an extremely broad field and growing every day. Over the past several years there have been numerous excellent presentations showing how data analysis, machine learning, and big data can be applied to derive new insight into petroleum systems. What we haven’t seen in these presentations is how geologists might leverage new technology to augment their capability and shift cognitive load to computer systems to greatly change information flow and decision making.

With that in mind, we have spent the last several years thinking about what geoscientists ACTUALLY do. In our estimation, the principal skill of geoscientists is applying domain expertise to sparse data. Something computers, no matter how sophisticated, can’t do. And what geoscientists ACTUALLY do is recognize patterns and find trends, mostly visually. Over the last hundred years, most advancements in geoscience are the result of new ways of displaying data and understanding relationships using visual correlation tools. Well known examples of visual correlation tools are Picket Plotts, Hingle Plotts, Buckles Plots, Horner Plots, DST pressure charts, generic cross plots, nomograms, cross section etc. And really it should probably be argued that the most humble tool in our kit, the well log, is nothing more than a tool of visual pattern recognition.

With this in mind, we set out to explore what we could accomplish using computer vision to teach a program to “see” like a geologist. In this talk we will present some results from a series of experiments on what patterns a computer can recognize in porosity-permeability plots and how these abilities can be leveraged to change what a geologists does. We hope you will join us for a discussion of cognition, geoscience and technology to think about what the future might hold for our profession.