Identification of complex salt-geometries from inaccurate migrated images using Deep Learning

What is it about?

A good velocity model is essential to produce an accurate seismic image in complex areas, and the inclusion of salt in the velocity models is still a challenge which consumes a huge amount of human and computational resources. This paper proposes a novel way of using Deep-learning to accurately predict the salt geometry from inaccurate migrated subsurface offset gathers.

Featured Image

Photo by Pietro Jeng on Unsplash

Photo by Pietro Jeng on Unsplash

Why is it important?

The proposed method does not require iterative salt inclusion, predicting the full salt geometry at once, even for complex cases. The technique has the potential to reduce the time and the risk of the conventional process.