Full text loading...
-
An Efficient RTM Method using Wavefield Compression with Combined Time-Space Dimensions
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 86th EAGE Annual Conference & Exhibition, Jun 2025, Volume 2025, p.1 - 5
Abstract
Summary
The reverse time migration (RTM) cross-correlation imaging condition requires to access the forward-propagated and backward-propagated wavefields simultaneously. To reduce the wavefield storage and I/O cost of RTM, we propose a time-space dimension forward wavefield compression and reconstruction method that combines both Nyquist time sampling and spatial wavefield random sampling. Firstly, in the time dimension, the Nyquist sampling law is utilized to reduce the sampling rate, hence only the forward wavefields at Nyquist sampling points are stored. Subsequently, in the space dimension, taking into account the sparse nature of the seismic wavefield, the wavefield snap at each Nyquist sampling point is sampled using the Gaussian random measurement matrix, thereby achieving compressed sampling of the wavefield. Before the cross-correlation operation, the ONSL0 algorithm is utilized to effectively reconstruct the wavefield. Experiments using a 2D layered model and a 3D gradient model demonstrate that our method effectively reduces the memory requirements. In addition, the wavefield reconstruction error using the K-SVD dictionary is smaller than the DCT counterpart, albeit at the cost of a slightly increased computational time. The proposed method can also be used in the gradient calculation in full waveform inversion.
© EAGE Publications BV