Your account already exists. Please login first to continue managing your settings.
Achieving a reliable post-stack seismic inversion represents an important objective in subsalt exploration.
Exploration for oil and gas below salt is fraught with technical and operational challenges. Operators need to determine the correct geometry and depth of subsalt reservoirs and predict distributions of porosity, lithology, fluid saturation, and other properties. How can they best achieve these goals?
The amplitudes of reflected seismic waves change at the interface between formations having different acoustic properties. Variations in acoustic impedance control these changes. As a result, it is possible to mathematically invert post-stack seismic reflection data to obtain the acoustic impedance of formations on both sides of an interface. By correlating inverted acoustic impedance with properties measured directly in boreholes, interpreters can extend known reservoir properties throughout the 3-D seismic volume between wells, reduce exploration risks, and lower costs. However, even advanced 3-D seismic reflection data suffer from three common problems below salt or near the flanks of salt overhangs. First, data quality is often poor because the signal-to-noise ratio is weak. Second, illumination typically is inadequate due to complex ray paths induced by salt structures. Third, traditional tomographic seismic velocities used for migration are almost always inaccurate or uncertain below salt. They tend to be of very low resolution and inconsistent with velocities measured in wells. Without reliable velocities, subsalt imaging is poor, and seismic inversions intended to distinguish rock and fluid properties below salt have proven largely unsuccessful.
As part of a major basin-to-prospect project in the deepwater Gulf of Mexico (GoM), WesternGeco and Schlumberger Data and Consulting Services developed an integrated geological and geophysical workflow to enable post-stack seismic inversions with sufficient resolution to characterize subsalt reservoirs. The workflow included three unique or uncommon components: (1) rock physics analyses from wells, (2) space-adaptive wavelet equalization (SAWE), and (3) integration of geology using a horizon-like layer sequence field (LSF).
Overview of the study area
Approximately 1,295 sq km (500 sq miles) of post-stack, wide-azimuth, anisotropically wave equation-migrated data were used for interpretation and inversion. Five seismic horizons were mapped below salt. Three subsalt wells were selected for post-stack inversion because they were widely distributed, and each had a complete suite of logs including compressional sonic and density curves. Two had checkshot data. A fourth was used for a blind test of the inversion process.
Integrated post-stack inversion workflow
The primary goal of the integrated post-stack inversion workflow was to investigate the reliability of seismic reservoir characterization below salt.
The workflow began with well log editing to remove spikes, cycle skips, noise, and bad or suspicious data. Sonic logs were calibrated with checkshot data to establish a time/depth function. Petrophysical analyses of all wells included estimations of acoustic impedance, clay volume, porosity, and fluid saturation.
Although density, gamma, and resistivity logs could discriminate between sands and shales, lithologic discrimination from calculated acoustic impedance was difficult. Compressional sonic logs were not reliable differentiators of either lithology or fluid content. Hence, rock physics analyses comparing computed density porosity, acoustic impedance, and clay volumes were conducted, successfully distinguishing between sand and shale facies. The use of rock physics was one of three key contributors to success.
The next step was to establish a well-to-seismic tie. A synthetic seismogram was generated from well data convolved with a wavelet extracted around each well. Vertical wells generate the most reliable synthetics. However, most deepwater wells are deviated or logged through reservoir sections that can include impedance anomalies. Therefore, initial estimated wavelets were chosen at each location through careful cross-correlation and validation between wells and inversion tests.
Due to overburden effects, estimated wavelets at each well location were quite different, although their phases were close to zero. This spatial variation implied that the extracted wavelets were not the same as in the 3-D migrated seismic volume. To correct the wavelets to minimum (zero) phase and stabilize any residual variation in space, the team used space-adaptive wavelet processing. The process incorporated SAWE, a unique component developed by Schlumberger, further contributing to success.
Signal-to-noise estimation (SNEST) also was carried out for each interval of interest. After SAWE, the team performed relative acoustic impedance inversion trace by trace using implicit spectrum inversion to remove the broadband zero-phase wavelet derived from SNEST. As expected, inversion showed that interpretations of seismic horizons below salt were uncertain since well and seismic velocities did not match. It was critical, therefore, to refine the seismic velocity and horizon picks prior to absolute acoustic impedance inversion.
Converting relative acoustic impedance to absolute acoustic impedance requires extremely low frequencies, often much lower than those contained in conventional seismic data. A low-frequency model incorporating impedance log data is necessary to properly scale the relative impedance model. As input to the low-frequency model, the team generated an LSF. This third unique component of the integrated post-stack inversion workflow estimates formation dips directly from seismic data along inline and crossline directions and converts them to a horizon-like scalar field. The LSF brought important geological information into the low-frequency model. Other inputs included well velocities, subsalt interpreted horizons, and the original migrated seismic velocity.
The LSF, horizons, and well velocities guided the extrapolation of acoustic impedance information away from well locations and refined the seismic velocity within the low-frequency model. The improved seismic velocity tied 100% at all well locations, and the low-frequency model now contained impedance variations below seismic resolution.
Finally, adding the refined low-frequency model to the relative impedance inversion scaled it to an absolute acoustic impedance volume that matched interpolated impedance logs from the wells. To ensure a good match away from the wells, the team employed structural interpretation and a weighted projection of the impedance logs onto each trace position in the 3-D volume. Results for a large inversion window showed that absolute acoustic impedance from inversion fit very well with calculated acoustic impedance at all three subsalt well locations.
Successful sand/shale prediction below salt
Petrophysical and rock physics analyses of edited logs established correlations between lithology and parameters from seismic inversion – acoustic impedance, velocity, density, porosity, and clay volume. These correlations aided qualitative interpretation and quantitative estimation of the sand/shale ratio. Comparisons of well lithology with predictions from the post-stack inversion workflow matched very well to the main Miocene and Wilcox reservoirs. In one well, for example, the clay volume predicted by inversion was close to the log classification. Although the blind test well was highly deviated, inversion results also were consistent with predictions from well logs.
However, the inverted porosity volume did not fit well with the porosity at well locations due to a lack of sufficient data. Only three deep wells were available for post-stack inversion. The team believes that if more well data were available, results would improve. Anisotropic reverse time-migrated data would probably yield better inversions as well.
This new workflow provided the first successful prediction of regional sand/shale distribution for subsalt reservoirs in the GoM. The approach can better help operators assess reservoir risk and serve as a guide for petroleum systems analysis. Finally, it can assist in refining pore pressure prediction to significantly lower sub-salt exploration drilling risks and costs.