Technical Abstracts

The paper proposes a method for adaptive subtraction of multiples in the wavelet domain using complex wavelet frame. The adaptation is performed for each scale of the transformed domain by computing a non-stationary unary complex filter that accounts for the phase and amplitude discrepancies between the model and the data. The method is extended to joint multiple model subtraction and tested on real data with good efficiency. Signal-to-noise ratio analysis on synthetic data allows choosing the adequate level of redundancy of the transform that is the best compromise between the quality of the subtraction and the computing time.

In order to investigate the impact on seismic inversion, comparative elastic inversion tests have been conducted using 2-D seismic data from conventional flat streamer and proprietary variable depth streamer acquisitions. In addition, a Bayesian lithology classification workflow has been performed on the inversion results to evaluate the impact on the prediction of lithology and hydrocarbon presence.

Recently, tilted transverse isotropic (TTI) imaging has become a standard practice in deep water Gulf of Mexico (GOM) to resolve the anisotropic effects of wave propagation in salt-withdrawal mini-basins. Tilted orthorhombic (T-ORT) anisotropy, which usually represents parallel-aligned fractures normally embedded in tilted thin sedimentary layers, is a less restricted assumption and can be applied to more complicated geological formations. This paper focuses on the key components of T-ORT depth imaging: T-ORT ray tracing, the procedure to build initial T-ORT models, and T-ORT tomography. With a real multi-azimuth (MAZ) dataset in the Perdido fold belt, GOM, we demonstrate the effectiveness of the T-ORT velocity model building, where the estimated T-ORT anisotropic parameters are consistent with the geological setting. T-ORT prestack depth migration flattens common image gathers (CIGs) in all azimuths and results in improvements on image focusing, as well as spatial positioning of complex structures in the area of interest.

In this paper, we demonstrate the procedure for T-ORT model building: the initialization of a T-ORT depth model and the implementation of T-ORT tomography. In order to show the effectiveness of the T-ORT model building with multi-azimuth data, we apply the T-ORT model building to two complex geology areas: a) in the Perdido fold belt, deepwater GOM, and b) a subsalt region in Green Canyon in central GOM. T-ORT model building in the fold belt case yields improved flatness in CIGs among azimuths and consequently enhances subsurface images in the area. In the subsalt case, we see improved imaging of the overburden mini-basins and, subsequently, improved salt geometry and better subsalt images with T-ORT RTM.

Angle domain common image gathers are recommended in Kirchhoff and reverse time migration for velocity model building in complex areas. For both these approaches there is a general agreement that the tomographic ray pairs are fully defined by the incidence and azimuth angle information and the reflection dip and that if the velocity model is correctly updated down to a given horizon, it is not necessary to shoot the tomographic ray pairs upwards through this horizon. We recall here, as shown by Montel and Lambaré (2011), that both these statements have to be mitigated when the common image gathers exhibit a significant residual move-out. We then discuss performing angle domain migration velocity analysis through accurately computing the tomographic ray pairs. Finally we demonstrate, with an application of non-linear slope tomography to a complex synthetic example, the critical role of a good understanding of the kinematic observed in the angle domain common image gathers.

Even if grid-based ray-based tomography has delivered continuous progress in terms of resolution and efficiency, it can still hardly recover strong velocity contrasts encountered in areas with salt bodies or chalk layers. The traditional solution for velocity model building in such a context is to perform a layer-stripping approach. In this approach, velocities and horizons are updated layer after layer recursively from top to bottom. Such a workflow is however time-consuming and prone to velocity errors being propagated into deeper layers as the model building progresses. We present here a solution to remedy these drawbacks. Our solution involves a non-linear tomographic approach combining information of dense dip and residual move-out picks and of picked horizons to update a multi-layer velocity model. While dip and RMO picks are used to update grid velocity globally within the layers by non-linear slope tomography, picked horizons are kinematically demigrated and remigrated recursively from top to bottom to relocalize major discontinuities in the velocity model. We present an application of the method to a real North Sea dataset with a comparison with the layer-stripping approach.

In 2010 CGGVeritas started processing a multi-survey dataset covering approximately 7000 sqkm. The project included several acquisitions with different wide azimuth cross-spread designs (receiver line and shot line spacing, orientation and bin sizes being different). Pre-stack merging such a variety of data often leads to acquisition imprints and noise in the migrated image, resulting in poor continuity and unreliable amplitudes. In order to minimize these artifacts 5D data interpolation was used with the goal of mapping all the different surveys to a single acquisition design. This resulted in a dataset with homogeneous sampling for the complete survey, thus allowing migration in the common offset vector domain. The 5D data mapping process fully preserved azimuth information, and resulted in a seamless survey giving the chance to have an ?artifact free? final image.