The inefficiency and inaccuracy of current methods in seismic imaging, or the use of vibrations in the ground in order to understand its composition, is one of the critical bottlenecks for appraising carbon capture sites, geothermal sites and natural gas reservoirs, amongst other applications, a process that is completely critical to the net zero transition. The carbon neutrality market alone is forecast (by the oil and gas industry itself) to be worth $100bn by 2050, involving 10,000 projects and the capture of 15m tonnes of CO2: this is likely a conservative estimate. Today, more than $8bn is already spent on seismic imaging annually, relying completely on months of manual annotation and interpretation by geophysicists.
The straight application of machine learning to seismic data would be extremely computationally expensive, and most likely would achieve only an incremental improvement. Optic Earth approach the challenge from a deep geophysics perspective and are focused on automated and instantaneous depth velocity modelling and image uncertainty estimation. Their vision is to build the world’s leading geophysics AI company, capable of seeing kilometres below the Earth’s surface in high resolution. Optic Earth are in paid pilots with some of the largest Energy firms in the world. They are manifestly a perfect example of an energy transition company, bridging natural gas, carbon capture and geothermal.