Joint inversion

Joint impedance inversion inverts both these data simultaneously, and as the recorded shear component is being utilized in the inversion, the process is considered superior to simultaneous impedance inversion. However, the success of joint impedance inversion depends on how accurately the PS data is mapped on the PP time domain. Normally, this is attempted by following well-to-seismic ties for both PP and PS datasets, and the matching of different horizons picked on both PP and PP data. Though, it seems to be a straightforward approach there are a few issues associated with it. One of them is the lower resolution of the PS data than the PP data which presents difficulties in the correlation of the equivalent reflection events on both the datasets. Even if, few consistent horizons get tracked, the horizon matching process introduces some artifacts on the PS data mapped into PP time. At SamiGeo, we follow a novel workflow for addressing them.

Well-to-seismic correlation for PS data as well as registration with PP data, at the location of a well. The PS synthetic seismogram (blue traces) is shown in (b) correlated with PS real seismic traces (in red). The displayed wavelet, used for generation of the synthetic seismogram, was extracted from the PS seismic data using a statistical process. The PS (c) and PP data (d) are shown in PS time. (Adapted from Chopra and Sharma, 2020)

Segments of seismic sections from (a) PP and (b) PS data in PP time. Four equivalent reflection events have been picked on the data volumes separately as seen by the blue horizons picked on the PP data and the magenta horizons on the PS data, but the horizon matching has not been done yet. The VP/VS values at every CDP are overlaid in colour. (c) The same PS section as in (b) but with the horizons (blue and magenta) matched. Notice, the revised values of VP/VS (which seem abnormal in the lower intervals) as well as the reflection distortions in the form of undulations. (Adapted from Chopra and Sharma, 2020)

Joint inversion workflow to address resolution and registration issues with PS data. (Adapted from Chopra and Sharma , 2020)

An arbitrary line section from (a) PP and (b) PS data in PP time after horizon matching. Four horizons have been picked on the sections and the impedance log curves have been overlaid. Notice in (b) there is some jitter seen on the left part, which is concerning. Also, the segment of the reflection event depicted in dashed black outline has a distinctly different character on the PS section, which prevents the imaging to this event on the impedance section. Both these observations signify artifacts. (Adapted from Sharma and Chopra, 2019)

An arbitrary line section from the P-impedance volume with four horizons and impedance log curves overlaid. Apparently, the inversion has not performed optimally. Notice the jitter on the left side of the section as well as the reflection event in black dashed outline in the above Figure carried through in the inversion. The mismatch between the inverted impedance and that measured in wells W1 and W2 (dark green dashed outline) as well as W4 and W5 can be seen clearly. (Adapted from Sharma and Chopra, 2019)

An arbitrary line section from the P-impedance volume generated using the proposed workflow, with four horizons and impedance log curves overlaid, which is equivalent to the section shown in the above Figure. The section shows much better correlation with the overlaid P-impedance log curves, and is free of artifacts. (Adapted from Sharma and Chopra, 2019)

References

• Sharma, R. K., S. Chopra and L. R. Lines, 2019, A robust workflow for performing joint impedance inversion with applications from North American Basins, published in Interpretation, T141-T149.

• Chopra, S. and R. K. Sharma, 2020, Addressing some artifacts in PP-PS registration prior to performing joint impedance inversion, published in The Leading Edge, 47-52.