The 12th International Symposium on Recent Advances in Exploration Geophysics (RAEG 2008)

During IODP Expedition 314, we used Schlumberger’s seismicVISION tool to monitor drilling depth progress in real time and to produce post-drilling depth corrections to our regional 3D seismic reflection data set. We collected data at four drill sites, including one site in the Kumano forearc basin that reached TD of 1401 m below seafloor, which we use as an example in this report.

We propose a newly-developed depth imaging approach for OBS (Ocean Bottom Seismometer)　reflection data in active-source structural surveys using wavefield separation and PSDM (Prestack Depth　Migration). OBS data includes a lot of valuable signals, not only reflection but refraction. However its wavefield is contaminated with various kinds of waves that degrade the quality of the depth imaging of OBS reflection. Multiple reflected waves, for example, have been considered as a strong source of noise in the imaging. However, we have found out that the multiples have a wide spread of reflection points and we take advantage of this feature to improve the depth imaging after careful processing of acquired OBS data. In our OBS survey, the spacing of OBS locations is typically 3-5 km. The sparse spacing means that signal-to-noise (S/N) enhancement by signal stacking is not expected. Therefore, we require higher quality data input to PSDM to obtain good depth image. OBS data contains upgoing and downgoing waves because it is located on sea bottom. We investigate the characteristics of OBS wave fields and classify them into four categories which include primary and multiple reflections. Each of them contains similar features in the signal, and this wave field separation approach can enhance the S/N ratio. Applying adequate PSDM to the four separated wavefields, we obtain the four depth images without contamination of the different type of wave fields.

This paper describes effective use of seismic attributes in the geostatistical reservoir modeling of oil sands. The goal of the work was to provide models for the reservoir simulation to forecast bitumen production from oil sands by Steam Assisted Gravity Drainage method (SAGD). Simulation models were required to reconstruct geological heterogeneity with sub-seismic resolution which can affect the steam chamber growth. Stochastic modeling approach was employed to generate realizations with the required resolution for uncertainty assessment. Depositional environment of the target formation was considered as fluvial-to-estuarine, and reservoir properties were varied drastically in both vertical and horizontal directions. Seismic data were used to capture the vertical and areal trend components in the reservoir property variations to take them into account for the application of the geostatistical tools. Correlation between seismic attributes and reservoir properties was examined to find that density and velocity provide useful information for both facies discrimination and mudstone volume estimation. The procedure of our reservoir modeling for oil sands was as follows. Firstly, multi-attribute analysis was conducted to obtain local facies probabilities. Secondly, sequential indicator simulation with locally varying proportion was employed for facies modeling, where the local proportion was provided from the local facies probabilities. Then, sequential Gaussian simulation with collocated cokriging was carried out for each facies to estimate mudstone volume distribution. Finally, effective porosity and permeability were estimated as a function of the mudstone volume.

A three-dimensional FEM (finite element method) for solving elastic or viscoelastic wave propagation problems are tested. Viscoelastic wave equations are introduced by three relaxation mechanisms. Second-order polynomial interpolation function and perfectly matched layer (PML) absorbing boundary condition are composed to this FEM framework. Also, it performs effective CPU time reduction by parallel computing. The accuracy of the FEM is evaluated by three tests. FEM is compared to the analytical solution for Lamb’s problem and the semi-analytical solutions for layered elastic or viscoelastic media. Especially, FEM and the analytical solution for Lamb’s problem showed almost perfect convergence. Therefore, we conclude that FEM produces very exact waveform. Moreover, this conclusion can support the accuracy of FEM for viscoelastic wave equations.