This reference is for an abstract only. A full paper was not submitted for this conference.
Seismic reservoir characterization (for instance, inversion) requires a well-controlled seismic signal that can sometimes not be guaranteed by conventional processing QCs. One option is to use borehole data to help choosing the processing parameters that will enhance the reservoir characterization reliability. This method is indeed very useful, but is limited since it only gives a very local estimation of the seismic quality at the location of the well. We thus propose in this paper a methodology based on much more general Quality Controls attributes derived from seismic at each key step of the processing. These attributes may be generated over the whole seismic survey and hence give a comprehensive idea of the seismic quality. This methodology has been tested and validated on a Tight Gas Reservoir field located onshore.
The field studied is an onshore gas field whose main objective is interpreted as late Paleozoic braided and meandering channels from fluvio-deltaic sequences, the thickness of which hardly exceeds a few meters. One of the main geosciences challenges is to define the lateral extent of these sand bodies: a 3D seismic pilot has thus been acquired to image this objective, with a nominal fold of 96 traces per Common Mid Point (CMP). Several wells have also been drilled in this area: a detailed Rock Physics analysis has been performed using the wellbore data acquired in these wells, and has proved the need to get reliable Acoustic Impedance (AI) and Poisson's Ratio (PR) to distinguish the sandstones from the other lithologies. In order to guarantee the integrity of the seismic amplitudes, and thus a successful Pre-Stack inversion, we decided to monitor the quality of the seismic at each key processing step, in terms of reliability for subsequent reservoir characterization. What is sought as a final product is a good quality Pre-Stack inversion, and in particular a reliable Poisson's ratio. To do so, we must at least ensure that:
1. The signal-to-noise ratio is as high as possible, while preserving a broad frequency bandwidth: high resolution, which is a function of S/N and frequency, is indeed paramount to detect the reservoirs;
2. All the substacks are consistent relative to one another, in terms of phase and time shifts, amplitudes and expected AVO behavior;
3. The absolute phase of the seismic data is well controlled, to allow further…
4. …meaningful and robust well calibrations. The extracted wavelets must be stationary (in amplitude, phase and spectral content), and representative of the whole seismic dataset.
Each of these topics has then been expressed as a specific Quality Control attribute, and has been computed on a trace-to-trace basis over the whole area of the 3D pilot, within three specific time windows (a first window centered on the objectives, a shallower one, and a deeper one), and at five milestones of the processing sequence. We were thus able to derive a rather global and exhaustive analysis of each attribute, which allowed us to get a more thorough understanding of the seismic data. As explained above, five general aspects of the seismic data quality were addressed and quantified during the whole processing sequence:
1. The lateral stability of the signal characteristics in phase: the dominant phase of the signal was computed along a particularly robust seismic marker on the five substacks and analysed in terms of statistical and spatial variations.
2. The frequency content of the seismic signal was assessed and monitored.
3. The Pre-Stack consistency, i.e. the way amplitudes vary according to the offset, and the resemblance between substacks were evaluated using different attributes. Cross-correlations along with time and phase shifts were computed between pairs of substacks. The variations of amplitudes with offset were assessed using intercept and gradient with the purpose of evaluating global mis-fit residual energy to a linear model across the various time intervals.
4. Amplitudes were of course examined with the classical RMS amplitude attribute. The RMS amplitude of the noise was evaluated, which allowed us to quantify the signal-to-noise ratio of the data, substack per substack, interval after interval.
5. Finally, borehole data (well log edited sonic and density logs, and corridor stacks) were used and integrated in this more global analysis.
Each aspect of the seismic data quality was quantified in order to obtain comparable results and normalized relative to the same quantity evaluated for the previous processing vintage, which we chose as a reference for the monitoring. The objective of the final processing was of course to improve the results so as to get increased grades for each category. As a whole, the results are very satisfactory: the final processing yields a new seismic whose signal is more stable along the offset direction, whose SNR is improved, and whose fourth and fifth substacks are particularly improved, which is a necessary condition for obtaining a reliable Poisson's Ratio from Pre-Stack inversion. On the field studied, Pre-Stack inversion of seismic data into acoustic impedance and Poisson's Ratio is paramount for a proper seismic reservoir characterization. Hence, seismic data processing has to be finely tuned to ensure a successful inversion, with in particular a good signal-to-noise ratio and an adequate substack consistency. A tight quality control methodology was thus put in place to monitor the seismic signal quality throughout the main steps of the final processing sequence. The latter involves both borehole data and more global seismic attributes. This methodology provided us with a precise insight into the seismic data quality from the very beginning of the processing sequence, and also helped us understand some key parameters (such as surface conditions) that before all impacted the seismic quality. The satisfactory results obtained show that seismic processing, when aiming at reservoir characterization studies, should be performed in close collaboration with the Seismic Reservoir Characterization.