(Press Release Courtesy of UL Lafayette)
Prior to The American Association of Petroleum Geologists' 2018 annual convention and exhibition (ACE), a subsurface hackathon was put on by Agile Scientific, a geology and geophysics consulting company based in Canada. The theme of this year's "subhack" was to combine machine learning and subsurface geology while defining an existing problem/hurdle in subsurface geology. Teams were tasked to come together over the three-day event to develop code that would help solve the defined problem. At the end of the event, geo-hackers presented their prototypes to a judges' panel that consisted of special guests from academia, industry, and sponsors of the event.
This year, Mark Mlella, a ULL geology graduate student, and his team won the subhack prize for the best execution for delivering a product that can solve an existing subsurface geology problem. The team developed a web application that predicts elemental capture spectroscopy (ECS) log from triple-combo logs. Machine learning was used to train a model by feeding triple-combo and ECS logs, in turn, the developed model was used to predict ECS by only using the triple-combo logs as the input parameters.