Search

We use the concepts of entropy and information theory to design a confidence measure for Bayesian facies estimations. Bayesian analyses provide the probabilities of occurrence of each constituent facies in a set. The entropy analysis uses all of these to establish a Confidence Index describing the reliability that the most-probable facies is in fact a clear best choice. We apply these ideas to various facies estimates from Gulf of Mexico inversions.

We address the problem of creating low frequency models for AVO inversions, investigating two workflows, neither of which involve well log interpolation. The first workflow creates a constant but structurally-compliant low frequncy model from averages over available logs. Facies analysis is done from the inversion outcomes. The second workflow takes the facies from the first workflow and together with per-facies trends creates a new model for a second pass of inversion. The results of the subsequent facies analysis are compared with the first in several ways and found to be superior. Notable in the analysis is the inclusions of several types of uncertainty which becomes useful in assessing risk.

CGG, together with TGS, recently acquired a multi-client TopSeis™ 3D seismic survey in the Barents Sea. The survey, known as “Greater Castberg” was acquired and imaged using the latest advances from CGG's proprietary TopSeis solution. The latest advances include a wide distribution of five sources over the spread together with one source in front of the spread. An early fast-track dataset shows outstanding imaging of the Greater Castberg area’s highly diverse and complex geology. An advanced fast-track volume will be ready for licensing in May and the final dataset will be available in Q1 2021.

We discuss the advantages of multidimensional (in data space) optimal transport (OT) full waveform inversion (FWI). We show that a careful formulation results into an enhanced coherency of the events continuity in the move-out direction, that leads to an improved velocity update compared to least squares but also to monodimensional (in data space) OT FWI. This is illustrated on both marine and land field datasets.

Paper designed to focus on new technology - our specific focus is on the use of modern processing sequences applied to 2D legacy data in Peru (GeoSpec). There are links drawn from previous geological study of the area.

Main authors are BP and CGG are co-authors so limited on what updates can be made. Ongoing thoughts are that CGG would like more acknowledgment for the "EVO"; approach but other than that happy with the content.

Shallow S-wave velocity model building is very important to land and Ocean Bottom Seismic (OBS) PS data processing. Surface wave, propagating along near surface, brings lots of useful information of shallow lithology which is good for near surface S-wave velocity estimation. In this study, multi-modal surface wave inversion (MM-SWI) was used to build high resolution shallow S-wave velocity model for one of the recently acquired Ocean Bottom Node (OBN) surveys. To generate high signal-to-noise ratio (S/N) dispersion spectrum for MM-SWI, super gathers based on the extracted surface waves were constructed. Aperture (within 400 m) was carefully chosen to ensure dispersion spectrum is of high spatial resolution as well. Auto-picking of multi-modal dispersion curves was implemented after the dispersion analysis. To overcome the strong nonlinear problem during multi-modal inversion, the combined Levenberg-Marquard (LM) and differential evolution (DE) inversion method was used. High resolution dispersion analysis and robust auto-picking of multi-modal dispersion curves enhance productivity significantly which allows a high resolution shallow S wave velocity model to be built efficiently for this survey. The inverted MM-SWI model correlates well with both the shallow geological structures and the seafloor map. It is able to reveal the fine layers, small faults, small channels and other lithological variations. PS statics was then computed from the inverted high resolution Vs model and applied to improve the continuity and coherency of the PS time image. The MM-SWI model was also incorporated into the S-wave velocity model building process for PS depth imaging. The evidences from the model comparison, statics application and PS depth image support the reliability of the high resolution shallow S-wave velocity model from MM-SWI.

The Mungo field in the North Sea is both structurally and stratigraphically complex. Despite improvements of OBC acquisition over the legacy towed streamer data interpretational challenges, particularly with respect to the salt geometry, remained. To achieve the required step-change in imaging within the project’s short timeframe, a high degree of technical content had to be adopted by employing a robust pre-processing sequence, including up-down deconvolution, along with an iterative approach to the velocity model build, utilising GWI, FWI, high-density tomography and anisotropic information derived from the PS data.

South Arne presents a velocity-model building challenge with large gas cloud obscuring underlying chalk reservoirs. Above the gas cloud is a complex near-surface with lateral velocity (Vp) variations. Additionally, a thin layer of strong anisotropy (peak epsilon around 23 %) occupies a 200-500 m interval in an otherwise isotropic overburden. The velocity model was updated with focus on reducing parameter cross-talk using 2-parameter Vp/epsilon joint FWI, then Vp/Q joint FWI. The FWI is run with combined OBN and towed-streamer input data, using a shot weighting strategy to balance the contributions to the gradient and global cost.