We investigate a recently proposed method for optimal sensor placement that we have adapted for a time-lapse seismic acquisition application. The premise of our paper is that a dense acquisition conducted for the base survey can be used to design an optimal sparse acquisition geometry for the monitor survey. The method uses Proper Orthogonal Decomposition (POD) to extract the basis from the training dataset. The base survey provides the training dataset to estimate the POD. Instead of using a conventional universal basis, the POD basis extracted from the base survey provides a data representation that is particularly tailored to seismic data reconstruction. We determine the optimal acquisition geometry for the monitor survey via dimensionality reduction and QR factorization with column pivoting. Pivoting permits to determine the optimal source and receiver geometry for the monitor survey. Once the optimal geometry is estimated, the least-squares fitting of the POD basis is used for reconstructing the monitor survey. We also provide a comparison between random sampling followed by reconstruction and the proposed method.
Presentation Date: Tuesday, October 13, 2020
Session Start Time: 8:30 AM
Presentation Time: 10:35 AM
Location: 362A
Presentation Type: Oral