Full text loading...
-
Wavefield Reconstruction Inversion With Enhanced Hybrid Tikhonov and TV Regularization for Imaging Piece-Wise Smooth Media
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 81st EAGE Conference and Exhibition 2019, Jun 2019, Volume 2019, p.1 - 5
Abstract
The subsurface may be described by piecewise smooth media. In such media, regions of smoothly varying properties and those with sharp transition zones have different statistical properties. This implies that hybrid regularizations may be necessary to reconstruct properly these properties by full waveform inversion (FWI). Recently, Hybrid Tikhonov + Total-variation (TT) regularization has been successfully applied in the framework of the improved wavefield reconstruction inversion method (IR-WRI), a promising extended search space approach which breaks down FWI into two linear subproblems: wavefield reconstruction driven by the recorded data through wave-equation relaxation and model parameter estimation by source residual minimization. Our TT regularization relies on infimal convolution where the smooth and blocky components are explicitly introduced in the model parametrization in contrast to the additive coupling of Tikhonov and TV regularizations. We first improve our TT regularization by jointly updating the smooth and blocky components of the subsurface by a variable projection approach compared to the original alternating-direction reconstruction. We show that the variable projection improves convergence speed, while being less computationally expensive. Then, we assess our new TT regularization against total generalized regularization, another regularization suitable for piecewise smooth media, based upon infimal convolution of first-order and second-order TV regularizations.