When it comes to building a scientific model of rocks buried kilometres underground, scientists and engineers need to work with huge volumes of data and run intricate simulation models. This talk describes how BP has grown a large community of Data Scientists, who are applying the open source data toolkit to transform and super-charge many of these traditional tasks.
Making an accurate model of rock formations which have been buried over millions of years is an enormous challenge, requiring the combination of geological intuition, physics-based models and massive distributed computing.
I will give a crash course in geoscience and introduce you to the methods used to describe the subsurface, combining physics-based and data-driven methods: from seismic surveys and gamma ray sensors to inversion modelling on the HPC supercomputer.
I’ll also highlight some of the subtleties encountered when applying machine learning methods to spatial data, such as devious information leakage.
Crash course in geoscience
Data science and machine learning in geoscience
Building a community of data scientists