Loading...
Seismic deblending: using iterative and compressive sensing methods to quantify blending noise impact on 4D projects
Velasques, Max M.
Velasques, Max M.
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2020
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Deepwater seismic acquisitions are well-known for their intensive use of financial resources, and the oil & gas industry continuously seeks for ways to acquire seismic data at lower costs to make large offshore projects viable. Blended seismic acquisitions (also known as simultaneous acquisitions) provide an affordable way to acquire seismic data. Blended seismic data makes deblending techniques an essential step in the seismic processing workflow because these methods allow for separate blended shot records, ensuring a high signal-to-noise ratio. Most applications of blended acquisitions rely on conventional 3D projects for exploration purposes, where the residual noise due to seismic interference is not a crucial problem when generating the final seismic image. However, for reservoir characterization, some uncertainties need to be analyzed when time-lapse blended seismic data is considered. In this project, I implement and compare two inversion-based deblending techniques for use with synthetic and field data: iterative deblending and deblending by compressive sensing. I also quantitatively compare these methods using 4D synthetic data to understand how blending noise and deblending procedures can influence the 4D data. I found that the tested deblending methods yield very similar results, although their need for computational resources is very different. The analysis with synthetic data before and after the introduction of random noise made the simulation more realistic and allowed me to understand how random noise marginally affects the results. Tests with field data confirmed the effectiveness of the deblending methods, even when missing traces and lateral amplitude variations were present. From a 4D perspective, under the modeled conditions, my results indicated that the NRMS magnitude (a measurement of repeatability) would be increased by less than 0.5%. This would be the size of the impact of a blended acquisition on our time-lapse project.
Associated Publications
Rights
Copyright of the original work is retained by the author.