Modelling continuity in seismic data

Seismic data is created by propagating a controlled shockwave into Earth's subsurface and interpreting the reflections of the shockwave. Different shifts in the geology of the subsurface will be detectable in the seismic data, which makes it useful for oil exploration. Visual Intelligence collaborates Equinor to develop models that can exploit seismic data and model the continuity of the subsurface.

Seismic data image the geology of the subsurface represented as a 3D cube. Hence, it is a representation of continuous surfaces and objects in 3D. Visual Intelligence and Equinor collaborates to find models that can exploit this type of contextual information and establish continuity in 2- and 3D. Models of this kind would be useful for seismic interpretation both as input to manual and more automated processes.

Our collaboration will start an activity related to modeling of continuity, investigating potential approaches that can be used to obtain continuity predictions without training samples. Continuity predictions can then later be exploited to identify and follow continuous structures in 2D and 3D. Initial approaches may build from optical flow and be tested on unlabeled seismic to evaluate potential and derive further methodological directions.


Further reading

Opening the black box of AI
January 19, 2021
Deep learning and AI models must become interpretable, explainable and reliable before they can be utilized in complex domains.
Performing objective measurements in ultrasound images
December 18, 2020
Exploiting limited data to perform objective measurements in ultrasound images of the heart.
New methods for automatic change detection in aerial images
January 19, 2021
A collaboration with Terratec to develop deep learning methods to automatically detect changes when updating an existing map database.