Inversion to Reservoir Properties

Estimating Reservoir Properties from Seismic Data: A New Approach

In the oil and gas industry, quantitative estimates of reservoir properties—such as porosity, lithology, and fluid content—are customarily obtained through seismic inversion. Traditionally, this process involves two stages: first, computing rock properties (such as P- and S-impedances and density) through elastic inversion; and second, inverting these rock properties to extract the reservoir parameters of interest.

Traditional Seismic Inversion Methodology

Stage 1: Elastic Inversion to Compute Rock Properties

Elastic inversion generates rock properties by minimizing the discrepancy between observed data and synthetic data. This synthesis is achieved by modeling relationships that incorporate:

  • Amplitude Versus Offset (AVO)
  • Offset-Varying Wavelet
  • Low Frequency Model (LFM)

The inversion software iteratively searches for the best rock property estimates that result in synthetic data closely matching the measured data. However, this mathematically complex process often involves non-intuitive parameters and a sensitivity to parameter changes that is not fully understood by many users. In practice, parameterization is frequently adjusted through an iterative process, comparing the inversion results with equivalent well-log measurements.

The Role of the Low Frequency Model (LFM)

The LFM plays a critical role when inverting to absolute rock properties. It supplies the necessary low frequency content—including the direct current (DC) component—that forms the base upon which relative rock property changes are superimposed. Derived from non-reflective data such as well-logs and seismic velocities, the LFM is considerably larger in magnitude than the relative changes measured by seismic data. Even slight inaccuracies in the LFM can lead to significant errors in the final results.

Estimating Reservoir Properties from Inverted Rock Properties

Various methods have been developed to estimate reservoir properties from the computed rock properties:

  • Qualitative Analysis:
    Uses cross-plots or multi-variate spaces to visually define clusters of seismic attributes that correspond to reservoir properties determined from well-logs.
  • Deterministic Model-Based Approaches:
    Apply effective media relationships to relate a specific reservoir property (e.g., porosity) to a corresponding rock property (e.g., impedance).
  • Empirical Regression Analysis:
    Involves fitting inverted rock properties to well-log or core data measurements of the reservoir property. The derived empirical relationship is then extended over the 3D seismic volume to generate reservoir property estimates.

In many cases, analyst expertise is paramount. Often, different geoscientists perform the rock property inversion and the reservoir property estimation, which can introduce additional uncertainty into the workflow.

A Methodology for Reducing User Input and Simplifying Parameterization

A new methodology is proposed to streamline this process, reducing the need for extensive user input while simplifying parameterization. The key modifications include:

Pre-Processing and Data Conditioning

  • Offset-Equalization and Phase Correction:
    The seismic wavelet is first offset-equalized and phase corrected during the data conditioning stage. This step is performed prior to computing the relative rock properties.
  • Integration of the LFM:
    When required, the LFM is incorporated at the stage of computing reservoir properties. This strategy diminishes the necessity for a strict, rigorous low frequency model and bypasses the estimation of absolute rock properties.

Linear Combination of Relative Rock Properties

Reservoir properties are ultimately computed as a linear combination of relative rock properties. The parameters of the linear equation are derived from well-log data using a least-squares regression analysis. In this analysis, two or more relative rock property parameters (for example, relative LambdaRho and relative MuRho) are combined to yield an estimate of the reservoir property. When applicable, the LFM can be introduced into the regression model as an additional property, particularly when it closely mimics the reservoir property of interest—for instance, using p-wave velocity from seismic data as the LFM when estimating total porosity.

Quality Control and Seismic Volume Estimation

Quality control displays, such as those illustrated in practical examples, provide a means for analysts to verify the accuracy of the linear relationship between seismic attributes and reservoir properties. Once the regression analysis is complete, the final step employs the established linear relationship to compute the reservoir properties across the seismic volume. Relative properties are obtained by integrating the reflectivities of the rock properties, and these cumulative results are combined to generate estimates that are reliable and accurate.

Conclusion

The proposed methodology offers a streamlined and effective approach for computing quantitative estimates of any reservoir or resource property that can be expressed as a linear combination of relative rock properties. By bypassing the direct estimation of absolute rock properties and relaxing the strict demands of a rigorous low frequency model, this method enhances the efficiency and reliability of reservoir characterization. Applications include the inversion of seismic data for effective porosity, brittleness, and even mineralogical attributes—opening new pathways for integrated subsurface evaluation.Embrace this innovative approach to seismic inversion and reservoir property estimation, and experience reduced user input, simplified parameterization, and improved confidence in your geophysical interpretations.