Full text loading...
-
Real Data Comparison of EPSI and Robust EPSI Including the Near Offset Reconstruction
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 77th EAGE Conference and Exhibition 2015, Jun 2015, Volume 2015, p.1 - 5
Abstract
Modern surface related multiple elimination techniques use sparsity of the (primary) Greens functions in order to constrain the inversion process.
In this paper we compare Estimation of Primaries by Sparse Inversion (EPSI; Van Groenestijn & Verschuur, 2009 ) and its refinement called robust EPSI ( Lin & Herrmann, 2013 ) on a number of real datasets. The EPSI algorithm uses a multitude of inversion parameters such as sparsity per gradient update and windowing in time to constrain the solution.
The robust EPSI method enforces sparsity via l1-norm minimization, essentially using only the target residual as a parameter. We extended robust EPSI to include reconstruction of the near offsets, which was missing in the original work.
We found that, for the datasets investigated, robust EPSI yields fewer artifacts and more reliable amplitudes in the region of the reconstructed near offset data.
This is the first real data application of the robust EPSI method to reconstruction of the near offsets.