Highly repeatable time-lapse seismic with distributed Compressive Sensing–-mitigating effects of calibration errors

TitleHighly repeatable time-lapse seismic with distributed {Compressive} {Sensing}–-mitigating effects of calibration errors
Publication TypeJournal Article
Year of Publication2017
AuthorsFelix Oghenekohwo, Felix J. Herrmann
JournalThe Leading Edge
Volume36
Number8
Page688-694
Month08
Keywordscalibration errors, joint-recovery method, marine, random sampling, time-lapse seismic
Abstract

Recently, we demonstrated that combining joint recovery with low-cost non-replicated randomized sampling tailored to time-lapse seismic can give us access to high fidelity, highly repeatable, dense prestack vintages, and high-grade time-lapse. To arrive at this result, we assumed well-calibrated surveys–-i.e., we presumed accurate post-plot source/receiver positions. Unfortunately, in practice seismic surveys are prone to calibration errors, which are unknown deviations between actual and post-plot acquisition geometry. By means of synthetic experiments, we analyze the possible impact of these errors on vintages and on time-lapse data obtained with our joint recovery model from compressively sampled surveys. Supported by these experiments, we demonstrate that highly repeatable time-lapse vintages are attainable despite the presence of unknown calibration errors in the positions of the shots. We assess the repeatability quantitatively for two scenarios by studying the impact of calibration errors on conventional dense but irregularly sampled surveys and on low-cost compressed surveys. To separate time-lapse effects from calibration issues, we consider the idealized case where the subsurface remains unchanged and the practical situation where time-lapse changes are restricted to a subset of the data. In both cases, the quality of the recovered vintages and time-lapse decreases gracefully for low-cost compressed surveys with increasing calibration errors. Conversely, the quality of vintages from expensive densely periodically sampled surveys decreases more rapidly as unknown and difficult to control calibration errors increase.

Notes

(The Leading Edge)

URLhttps://www.slim.eos.ubc.ca/Publications/Public/Journals/TheLeadingEdge/2017/oghenekohwo2017hrt/oghenekohwo2017hrt.html
DOI10.1190/tle36080688.1
Citation Keyoghenekohwo2017hrt