Search

A robust surface wave inversion has been developed to invert near surface S wave velocity for depth model building and shear statics correction. The algorithm utilizes mainly two most effective inversion engines: the non-linear least square Levenberg-Marquardt (LM) inversion and the differential evolution (DE) method. Both of them can minimize the dispersion curve distance for simpler cases, or directly solve the Eigen-determinant of the surface wave secular functions for more complex cases. The algorithm has been applied to an OBC 3D4C data from off shore Vietnam, where the near surface geology changes greatly from rapid depth varying reef structures to floodplain alluvial fans. The surface wave, it is the Scholte wave in this case, has demonstrates variety of dispersion patterns, which are used in the inversion and revealed the complex subsurface velocity model underneath the sea floor. The receiver side shear statics has been calculated based on the inverted Vs model and applied to the PS wave data, which not only saves man powers tremendously compared with the conventional shear statics method, but also provides a more accurate solution. The imaging results show much better event continuity and geologically interpretable horizons.

Most deblending approaches utilize coherency criteria in a certain domain to separate blended signals. Due to the presence of noise, such separation in the time domain can never be perfect. Conservative deblending is often performed in order to protect primary signals. As a consequence, residual source cross-talk noise can remain in the data after deblending, which could affect the subsequent pre-processing steps such as de-ghosting and de-multiple. Here, we present a wave equation-based demigration flow to help protect primary signals for deblending. It can be combined with existing deblending approaches to achieve clean deblended results with minimal primary damage. The workflow is tested on a numerically blended data set using acquired wide azimuth data. We also demonstrate its ability to overcome deblending challenges when the shots are coarsely sampled, and the dithering time is small.

OBN SRME that combines OBN and streamer data is known to be an effective way to predict surface-related multiples in OBN data. However, the available streamer data often have limited offset/azimuth coverage. Additionally, the double source wavelets due to the cross-convolution of OBN and streamer data limit the bandwidth (loss of low and high frequency) of the multiple prediction. OBN model-based water-layer demultiple (MWD) overcomes such limitations and is a good complement of OBN SRME; MWD replaces the streamer data with the water-bottom Green’s function that has no offset/azimuth limitation and keeps the full bandwidth of the input data. With Gulf of Mexico (GOM) OBN data over the Atlantis field, we illustrate the benefit of joint SRME and MWD over solely SRME with the improved attenuation of low-frequency multiples and deep peg-leg multiples.

Offshore development requires detailed understanding of subsurface reservoirs and proper well placement. Use of all available information in an integrated reservoir model leads to improved production rates and higher EUR. Conventional methods use well logs, geologic information, and structure from seismic interpretation in static reservoir models, which, in turn, predict future production and evaluate alternative management scenarios. However, these models, which use different data types in isolation, often fail replicate past production (history matching). An integrated seismic-to-simulation workflow is presented in this paper; dense 3D seismic data provide a great deal of lateral lithofacies and rock-property information that is essential to the static model.

Azimuthal inversion is state-of-the-art inversion technology for stress and fractured reservoir characterization and detection in anisotropic media. This technology requires a wide-azimuth seismic survey and careful azimuth dependent processing with noise attenuation. In this paper, the influence of noise attenuation on the reliability of anisotropic inversion results is discussed. The implementation of this technology for the understanding of horizontal stresses and the verification of the results with well data and microseismic will be presented.

The search for hydrocarbons has spanned vast areas of the globe. Despite many successes, entire basins and plays have been deemed unprospective, perhaps prematurely, based on single well results, or poor-quality seismic data. The advent of broadband seismic acquisition, combined with new deghosting/imaging algorithms and exponential increases in computer processing power, has generated improved seismic images. These advances give scientists an opportunity to re-evaluate a region’s subsurface geology. The new interpretation, combined with geology-based datasets, is now being studied in conjunction with modern and historical seismic data, to fully understand petroleum prospectivity.

We demonstrate a 4D FWI technique using a synthetic study involving 3D elastic modeling through a highly realistic Earth model akin to the actual Grane PRM data. For this acquisition configuration, this study also indicates there is minimal sensitivity of the method to various residual uncertainties in the data and the modeling. 4D FWI results using the real time-lapse Grane PRM data with a six month acquisition interval between vintages shows changes at the reservoir level that correlate with both injecting and producing wells. In addition, we find good agreement comparing the velocity differences from 4D FWI to 4D time-shifts from the fully processed and imaged seismic data.

This paper reviews elastic property changes in anisotropic mediums and proposes a workflow to model elastic stiffness tensor using conventional well logs in a vertical transverse isotropic medium. The proposed workflow uses the concept of downscaling and then upscaling of normal logs using Backus model and rock physics modeling which couples different effective medium theories together. The choice of the rock physics model depends on the anisotropy source in the given medium. Rock physics modeling is the main step in this workflow, and is very important in defining the final elastic constants of the medium. It is the place where all anisotropy information is integrated with the log information. Finally, this workflow is tested on the Duvernay shale formation in the western Canadian sedimentary basin where log data along with the laboratory core measurements are available. The good match between modeled and measured velocity confirms the validity of this workflow.

The Northern Viking Graben seismic dataset was acquired using CGG’s BroadSeis™/BroadSource™ true broadband solution, which combines a multi-level source and curved variable-depth streamer technology with advanced and customized imaging technologies. The dataset covers more than 35,000 km2, spanning both the Norwegian and UK North Sea. The high-resolution seismic imaging is a key component of the ongoing integrated studies, which will enhance subsurface knowledge and assist with the identification of prospects and new play models.