68th EAGE Conference and Exhibition incorporating SPE EUROPEC 2006

It is now commonly acknowledged that spatial filters using geostatistics (factorial kriging) may prove useful when applied to seismic cubes in several operational contexts such as spatial quality check, noise reduction, 4D equalization and spectral decomposition. But even when based upon a reliable co-regionalization model defined along the main inline, crossline and time or depth axes, these spatial filters are opened to criticism as no structural dipping information is taken into account. A dip steered factorial kriging solution is presented which consists in introducing local dip and azimuth information into the kriging equations system, thus leading to more geologically consistent results.

A real data case study illustrates the contribution of the dip steered factorial kriging when applied to a post-stack amplitude cube for noise reduction purpose: The coherency of the geological seismic events is enhanced when compared to results obtained by factorial kriging disregarding structural dipping.

Surface-consistent residual static correction is applied to land data on a routine basis. However, if the statics are large, conventional methods often break down due to, among other reasons, the difficulty to construct a good pilot trace. We developed a large static correction method that constructs the pilot trace by a local robust L1-norm inversion at each CDP location. The local inversion calculates the relative time shifts of the traces within a CDP gather by minimizing the trace-to-trace time difference. A pilot trace is obtained by stacking the CDP gather after aligning the reflectors by applying the time shifts to the traces. Finally, a global L1-norm inversion resolves the source and receiver statics from the time shifts of traces of all CDP gathers relative to their pilot traces. The method is demonstrated with synthetic and real data examples.

Long offset analysis of seismic data has gained interest during the past years, both for conventional AVO-analysis and time-lapse seismic data. A common problem for such analysis on marine seismic data is the presence of overburden generated refraction noise, similar to the ground roll noise problem of land data. A synthetic study shows that it is possible to diminish this overburden noise by beam-steering the source array. In marine seismic acquisition this is a well known technique, and we find that for a synthetic modeling example the beam-steering technique might attenuate this type of noise by a factor of 2-3, possibly enabling a better utilization of long offset seismic data.

A new parametric method for phase velocity estimation with an array of sensor is presented. It does not need a preliminary group velocity calculation and it estimates the function as a whole, providing a continuous and analytic expression. The novelty of the method stands in the velocity model and the initialization of the estimation. The model is based on a spline. This makes the functions smooth enough to guarantee the method's robustness to noise and undesired waves. The search of the best velocity function is initialized with a rough estimate provided by a frequency-wavenumber transform of the data, which highly helps the convergence. This method has been applied to synthetic and real surface wave signals. In both cases the signals corrected with the estimated velocity have a good alignment, in nearly all the spectrum of the wave.

We propose a data reconstruction based on a combination of migration Kirchhoff-style algorithm that migrates coefficients obtained by curvelet decomposition of seismic traces over time and a demigration scheme that generate data to the desire location. Curvelet data reconstruction involves four steps: Curvelet decomposition of the seismic data, thresholding of the resulting curvelet coefficients, multiscale Kirchhoff like migration, and demigration or data reconstruction from the multiscale images.

The migration procedure applied to each curvelet scale is the same as conventional Kirchhoff migration but operates on curvelet coefficients. Since only the curvelet coefficients are migrated, the cost of curvelet-based migration is reduced compared to that of conventional Kirchhoff migration.

Marine surveys acquired in deep-water areas often exhibit static variations between sail-lines that are the result of changes in sea level elevation (due to tides) or/and in water velocity. They manifest themselves as lateral discontinuities (or jitters) on cross-line sections or on 3D CDP gathers and can lead to stack deterioration. In this paper, we review the methodologies available to correct for these two types of variations. We will first show how GPS data can be used to compensate accurately for tidal statics. We then propose to correct for water velocity variations using a direct measurement on seismic gathers. The objective is to replace the measured water velocity (varying from sail-line to sail-line) by a constant water velocity that is consistent across the whole survey. This is the key point of the methodology as all the sail-lines (and all the vintages in 4D) will be re-aligned to this spatially constant water velocity. The main advantage of this method is that it does not rely only on a water bottom time measurement, which could be biased by residual tidal statics.

Extended Elastic Impedance (EEI) has proved to be a highly convenient framework for seismic AVO studies. EEI logs can be directly related to the petrophysical properties of interest and seismic EEI reflectivity volumes can be obtained directly from the prestack data via linear projection in sin2theta. The directness of the EEI method is one of its strengths, making it unnecessary to perform intermediate steps such as examining conventional AVO attributes (e.g., intercepts and gradients). Consequently it can be difficult for those more familiar with the conventional AVO analysis to understand the parallels between the two methods. Once these parallels are understood we show how they can be exploited to provide new AVO attributes for background Vs/Vp ratio and AVO class based on the EEI seismic projection method.

We propose a technique for determining elastic properties of a medium by the AVO-inversion for postcritical reflections. The inverse problem is solved by Nelder-Mid optimization method. Numerical experiments were used for testing the accuracy and stability of the algorithm for different offset ranges and initial models. It is shown that the velocity estimations can be obtained with the error <1% for all the offset ranges considered, if noise is absent. In the presence of noise the accuracy of the S-wave velocity estimation is lower at the precritical offsets. Using the nearcritical and postcritical offsets we obtain the errors <1% in the P-wave velocity and <6% in the S-wave velocity , when signal/noise ratio is >5

Over the past decade, time-lapse (4D) seismic has evolved into a valuable reservoir-monitoring tool used on numerous fields throughout the world. 4D seismic has been used at the Ekofisk field to monitor water injection and reservoir depletion. The 4D seismic signal is complex and can be broken down into two main components. The first and most apparent of these being 4D travel time differences caused by the geomechanical effects of compaction and stress changes occurring in the reservoir and overburden. The second and subtler component of the 4D signal is an amplitude difference caused by impedance changes occurring as the reservoir responds to water injection and pressure depletion. The combined response from these two 4D attributes has been used to optimize well placement and reduce the risk of drilling into water swept areas of the reservoir.

The 4D ROVER (Remaining Oil Volume Estimation And Redistribution) project aimed to provide a quantitative measure of remaining oil in the oil rim from the Troll West field, by deploying the Shell proprietary probabilistic layer-based inversion tool to the license partners. Due to structural complexity of the static model, the three reservoir fluids and a complicated 4D signal, a stepwise inversion approach proved most successful, solving for the most uncertain reservoir parameters first.