Subsurface velocity inversion from deep learning-based data assimilation

• Published in: Journal of applied geophysics 2019-08, Vol.167, p.172-179
• Main Authors: Mao, Bo, Han, Li-Guo, Feng, Qiang, Yin, Yu-Chen
• Format: Article
• Language: English
• Subjects: Data assimilation; Deep learning; Full waveform inversion; Prior velocity information
• ISSN: 0926-9851; 1879-1859
• DOI: 10.1016/j.jappgeo.2019.04.002

Data assimilation has widespread applications in meteorology and oceanography. This paper applies data assimilation for seismic exploration. Comparing with full waveform inversion, the distrust of the prior velocity information is preserved, to avoid the local minima caused by its inaccuracy. Prior velocity information is scarce in seismic exploration, but the introduction of deep learning methods makes it possible, to correct the major defects in the application of data assimilation in seismic velocity inversion. Finally, we design a simple salt body model, and we introduce deep learning to obtain prior velocity information and drive data assimilation for high-precision inversion of subsurface velocity. It is compared with the traditional full waveform inversion, to prove the effectiveness and advantages of this new process. •DL can build prior velocity information to mitigate the cycle skipping problem, even without low frequency information.•Data assimilation can avoids the local convergence caused by inaccurate prior velocity information.•Tensorflow developed by Google enables deep learning to run efficiently.