Velocity model building by deep learning: From general synthetics to field data application

Abstract
Velocity model building is not straightforward in geologically complex environments. We train a convolutional neural network (CNN) to map full wavefields to smooth subsurface parameter distributions to address the problem. Specifically, cubes of neighboring CMP gathers are mapped into in 1D vertical profiles to simplify the training phase and to make it easier to utilize well logs in future applications. We train the CNN using a total of one hundred thousand random subsurface models generated on-the-fly and the corresponding synthetic data. The application of the trained CNN on synthetic and real data admitted reasonably accurate models representing mostly the low wavenumber features of the true models.

Citation
Kazei, V., Ovcharenko, O., & Alkhalifah, T. (2020). Velocity model building by deep learning: From general synthetics to field data application. SEG Technical Program Expanded Abstracts 2020. doi:10.1190/segam2020-3428324.1

Acknowledgements
We thank Saudi Aramco for its support and CGG for sharingthe broadseis field data. The research reported in this publication was supported by funding from King Abdullah Universityof Science and Technology (KAUST), Thuwal, 23955-6900,Saudi Arabia

Publisher
Society of Exploration Geophysicists

DOI
10.1190/segam2020-3428324.1

Additional Links
https://library.seg.org/doi/10.1190/segam2020-3428324.1

Permanent link to this record