Search

Existing methods for addressing cycle skipping in full-waveform inversion (FWI) typically involve either a modification of one of the data sets used to compute the least-squares objective function, or a reformulation of the objective function itself, often in terms of a traveltime (or equivalent) misfit. Both approaches can be successful, but they are reliant to varying extents on the notion of event similarity – that is, the requirement that the observed and modeled data contain the same, distinct, seismic events, even if the corresponding kinematics are different. We introduce a new technique for mitigating cycle skipping in FWI based on partial matching filters. The method accommodates amplitude differences between observed and modeled data, and does not require any major modification to an existing inversion engine. The proposed approach is validated on synthetic and real data sets, including an example where we observe a reduced reliance on event similarity compared to an established cycle skipping mitigation technique.

With a shallow anhydrite layer, strong multiples and converted wave contamination, Southern Oman represents an outstanding challenge for land velocity model building and imaging. While acoustic land full-waveform inversion (FWI) has proved successful on new broadband datasets in Northern Oman, no successful application has been reported for Southern Oman. We show here that the challenge of acoustic FWI in South Oman can be overcome using a dedicated workflow combining Multi-Wave Inversion (MWI) and multi-Dimensional Optimal Transport FWI (multiD OT-FWI). The key component of the workflow is the very near surface characterization provided by surface wave dispersion curves, which allows delineation of the Rus layer in the initial FWI model. MultiD OT-FWI is then used to mitigate amplitude issues in the presence of short period multiples and reduce cycle skipping beyond the depth of penetration of diving waves.

We present a 4D case study of time-lapse multi-well Vertical Seismic Profiling (VSP) acquired with Distributed Acoustic Sensing (DAS) at the Mars field, deepwater Gulf of Mexico. In this work, we are able to obtain meaningful 4D signals from multi-well DAS VSPs by addressing various challenges that include weak 4D signals from a relatively short time interval (~1 year), an extremely noisy monitor survey, and poor repeatability between baseline and monitor acquisitions. 4D friendly premigration co-denoise was the key to attenuating the strong background noise in the monitor survey while preserving the potentially weak 4D signals. 4D shot co-selection and regularization effectively mitigated some of the acquisition differences between the baseline and monitor surveys. In addition, least-squares Kirchhoff (LS-Kir) migration compensated for the illumination variations and attenuated strong migration swings caused by irregular VSP acquisition geometry; it also improved the amplitude consistency between upgoing and downgoing wavefields within each well and among different wells. This facilitated the combination of three nearby wells with both upgoing and downgoing wavefields for 4D imaging and improved the coverage and S/N of the 4D results.

The use of graph-space optimal transport within foil-waveform inversion (FWI) has recently been proposed to enhance robustness to cycle-skipping. In this paper, we discuss several features of the graph-space optimal transport FWI, emphasizing in particular its ability to detect time shifts between modelled and observed data and its characteristics in terms of adjoint source (the signal back-propagated from the receiver side during the iterative optimization process). Our analysis is driven by finding an optimal adjoint-source that is robust to cycle-skipping. This leads us to propose an alternative graph-space-inspired scheme called enhanced kinematic transform FWI. Our approach involves a modification of the adjoint-source introducing a partial shift strategy, allowing to highlight the kinematic information. We illustrate the enhanced robustness with respect to cycle-skipping on synthetic and field datasets with comparison to conventional FWI, Kantorovich-Rubinstein optimal transport FWI and graph-space optimal transport FWI.

Complex overburden geology often creates difficulties in imaging reservoirs situated below. We used a field dataset from the GOM to demonstrate the benefits of combining FWI and LSM. We were able to generate a high-resolution velocity model through FWI, then use LSM to further improve the reservoir amplitude fidelity by compensating for the overburden and acquisition illumination effects through a round-trip modelling/migration process

The North West Shelf, situated in Western Australia, is a world-class offshore hydrocarbon province. The presence of hard water bottom, near water bottom reflectors and shallow Tertiary carbonates not only generates strong multiples but also distorts the ray-paths for deeper reflectors. Seismic data quality is severely deteriorated due to residual multiples, limited bandwidth, and poor signal-to-noise ratio, impeding reservoir delineation and further AVO/QI analysis. With the recent advancement of seismic imaging technologies, we propose an integrated workflow including (1) comprehensive demultiple and (2) hybrid tomography and Time-lag Full-Waveform Inversion (TLFWI) to overcome these long-standing imaging challenges. The significant uplift of the reprocessed image provides deeper insights into the subsurface geology and improves confidence of prospect mapping for exploration.

In the North Sea, water bottom related surface multiples offer one of the largest challenges to a seismic processing project with a careful balance between multiple attenuation and primary preservation required when using a model and subtract approach. For Ocean Bottom Node (OBN) data Up/down deconvolution can be used to output the reflectivity where multiple and source signature will be removed, eliminating the need for an adaptive subtraction as part of the principle de-multiple process. Residual multiple can however still be present where the assumptions within the up/down deconvolution method do not adequately describe the reality of the data. Often, these residuals are only noticeable once migrated, particularly where a diffracted component is prevalent. In this paper, a method using the coherency of residual multiple evaluated on migrated and stacked data is used to drive pre-migration residual multiple attenuation through the use of a matched-subtraction.

Stratal geometries, reservoir and source facies are different in true lakes, in lakes with restricted access to the global ocean, and in “lakes” continuously connected to the global ocean but capped by large quantities of freshwater. We illustrate the importance for exploration and development of addressing these issues in the pre-salt sequences of the South Atlantic. We have achieved this through interpretation of high quality seismic images of the pre-salt section, in a geological context provided by analysis of well and rock data, prediction of palaeoclimate and palaeohydrology, and an understanding of relative plate motions and global oceanic sea levels. We conclude that undrilled reservoir facies are present on the flanks of the main structural highs in the hypersaline basins of south Brazil, and in the freshwater basin of south Gabon.

Small-scale heterogeneities create distortions in the wavefield that are usually resolved by Full Waveform Inversion (FWI). When FWI fails (for lack of diving waves or low frequencies), recent works promote the use of targeted stochastic approaches maximizing the stack power. We instead revisit here ray-based tomography and propose an innovative approach for a high definition tomography which can recover small-scale anomalies without any a priori information on their localization. While previously proposed tomographic approaches aimed at vertical resolution and structural conformity, our new approach focuses on enhancing lateral heterogeneities. The first key component is an accurate RMO picking with structural dip information honoring the variation of dip with offset. Second, is an efficient nonlinear slope tomography employing small grids and a minimum level of constraints. Our approach is demonstrated on a 2D synthetic dataset with small-scale lateral variations and confirmed on a 3D streamer dataset from the Barents Sea.