Let's hear it for our Summer Interns!

Commentary from Porsche Adams and Man Fung Yeung
17 January 2022

This summer, we were delighted to welcome two student interns into the Earth Science Analytics (ESA) team. Between June and August, Man Fung Yeung and Porsche Adams, two Master of Petroleum Geoscience students from Imperial College London, utilised EarthNET software and the power of Machine Learning (ML) to conduct their respective thesis studies.

The internships were offered as part of the master’s study programme at  Imperial, with both students having been recommended to join ESA because of their interest in ML, coding, and data science. While they had basic experience in coding and strong geoscience backgrounds, they had never come across ML integrated software before, or used it in their personal or academic lives. Both reflected that at the outset, ML had appeared a  potentially inaccessible concept and practice for non-experts, with heavy code writing and backend requirements.

Despite initial reservations, Porsche and Man Fung threw themselves into the learning process with extensive hands-on practice with the EarthNET software. The ESA team were readily available via Slack messaging to offer their expert advice and guidance on how to use the software and extract the greatest value from their data. This meant that despite the restrictions on in-office working, Porsche and Man Fung still had access to the ESA team via weekly strategy meetings and regular chats with key experts including our co-founders Eirik Larsen, Behzad Alaei and Dimitris Oikonomou.

With the support of the team, both interns were able to get up to speed with EarthNET within the first three to four weeks, learning how to input data, QC and extract property predictions. Man Fung referenced this initial learning process as complex but well guided with the help of EarthNET’s built-in user guide. Similarly, Porsche highlighted the learning curve involved with transitioning to ML workflows but stressed that once picked up, these are skills for life that reduce future time and process burdens.

Once familiarised with the software, both students began work on their respective studies.

Man Fung explored the characterisation and detection of Paleogene injectites in the North Viking Graben, while Porsche was focused on overpressure mapping in the North Sea using deep learning techniques applied to fault interpretation. Both used data accessed through EarthBANK, and provided by companies such as CGG, and operators including Lundin Energy, to develop contextual knowledge and input into EarthNET for analysis. With integration from a range of datasets and software including Petrel, the students were able to draw on a huge wealth of data for their subject areas and work it into digestible insights.

For Man Fung, EarthNET allowed him to find the well and seismic data for the North Viking Graben quadrant 30-31, identify injectites, and generate 3D property prediction models of acoustic impedance using ML to predict their structure. With the use of ML workflows, Man Fung believes he significantly reduced the time that would be required to conduct a similar study through conventional means by streamlining the verifications and exact equations required to provide determinable results.

Porsche focused on training deep learning models to predict faults within the Jurassic Brent Group sediments and applied repeat formation tests and wireline log analysis to determine whether faults acted as pressure cell boundaries. The use of ML in her study enabled her to see the bigger picture compared to often incomplete manual methods, with the generation of a more refined fault network over the Brent Group. EarthNET also ultimately enabled her to conduct an intensive study in the short time available.

So, what was the result of their many weeks of hard work? Two detailed, knowledgeable, and insightful master’s theses, with findings generated through EarthNET.

It would be safe to say that both Man Fung and Porsche are now sold on the revolutionary nature of ML software practises, with both expressing an interest in exploring ML further in their careers. Porsche referenced the fact that our world is becoming ever more data-driven, with extraordinary amounts of data being created daily. As such, ML has the potential to revolutionise the ways we analyse, manage and use our data going forward, not just in the energy industry but across the board. She summed this up simply, stating: “Machine Learning is the future.” We couldn’t agree more, Porsche.

We wish Man Fung and Porsche the best of luck as they complete their studies and embark on their careers.