“Machine learning is particularly suitable for near-field exploration,” claims Eirik Larsen, founder and Chief Solutions Officer of the fast-growing company Earth Science Analytics, which has set itself the goal of building the “next generation workflow” using new computer technology.
The explanation is simple. Enormous amounts of freely available geological and geophysical data in mature areas of the Norwegian continental shelf, collected over almost 60 years, have not been used well enough because geologists – neither individually nor collectively – do not have large enough capacity to analyse it all. This applies to all types of data, from both seismic and boreholes, as well as countless images, analyses and reports.
Many will certainly recognize themselves in that description, and this is exactly what Larsen took on when he was faced with insurmountable amounts of information for use in both prospect evaluation and regional interpretations a few years ago.
“By using the computers’ ability to crush almost infinite amounts of data, completely new opportunities have opened up to gain insight into geological aspects that geologists previously could only dream of.”
In a few minutes, the machines can do what geologists will need for decades. In addition, “it is human to make mistakes”, and geologists are often biased and can therefore make mistakes.
New working methods and smaller organizations have also led to fewer experts doing the job. Machine learning can compensate for this.
“For the Norwegian Continental Shelf, it also means that even small companies with a limited staff can quickly build up a prospectus portfolio,” Larsen adds.
Instead of spending weeks and months on a particular exercise, machines will do the job in a matter of minutes, hours or days. For geologists, the advantage will be that they – after the machines have done the hard work – can spend more time on analysis, as well as make decisions based on large data sets.
“In short, machine learning helps to shorten the time spent compiling data, increase the volume of data that is actually analysed, improve the ability to integrate data types and disciplines, and thus the quality of geologists’ interpretations,” concludes Larsen.
Larsen is among the speakers during the DIGEX 2022 conference which this year will be held in Stavanger, as well as digitally, 6 – 7 April.
