Full text loading...
-
Downsampling Plus Interpolation for Wavefield Reconstruction by Reverse Propagation
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 78th EAGE Conference and Exhibition 2016, May 2016, Volume 2016, p.1 - 5
Abstract
Summary
In reverse time migration and full waveform inversion applications, one need to access both source and receivers wavefield simultaneously, which is computationally and memory intensive. Due to slow disk reading with huge volume of data, computation-based wavefield reconstruction methods such as checkpointing strategy and wavefield reconstruction by reverse propagation are preferred to achieve efficient implementation in this key step. Compared with checkpointing, reverse propagation is usually more efficient while suffering a stringent memory bottleneck for 3D large scale imaging applications.
In this study, we show that the heavy boundary storage can be dramatically reduced up to one or two order of magnitude, based on the temporal sampling determined by Nyquist principle, rather than the more restrictive relation from the Courant-Friedrichs-Lewy condition. We present three boundary interpolation techniques, namely the discrete Fourier transform interpolation, the Kaiser windowed sinc interpolation and Lagrange polynomial interpolation. The three interpolation methods, in conjunction with different computational efficiency depending on the global (Fourier) basis or local (windowed sinc and polynomial) basis, allow us to accurately reconstruct the boundary elements without significant loss of information, making the in-core memory saving of the boundaries practically feasible in 3D large scale imaging applications.
© EAGE Publications BV