Low Frequency Model Update for Thin-bed Coals - Integrating Spectral Decomposition and Inversion for Enhanced Imaging

This study elaborates on how to use a spectral decomposition technique to improve the quality of the seismic inversion result in media containing local stacks of thin beds. In seismic inversion, the band-limited part of the inverted result is obtained directly from the seismic data, while the low-frequency part is supplied by a trend model interpolated from well control. In thin-layer media the well control is too sparse to capture the rapid lateral variations in the low frequency trend. In this paper, a practical methodology is proposed to reduce these artefacts. This is achieved by: 1) identifying thin beds using spectral components of the seismic data and 2) updating the elastic parameter of the basic trend model at the locations of the captured thin layers using results from a first-pass seismic inversion. The applied updates reduce the effect of low P-impedance artefacts. In the examples shown here the artefacts are in a form of side-lobes caused by thin coal beds. The proposed technique is shown to work on a synthetic example and successfully applied in a real case study. An important advantage is that the method does not require an accurate geological model, but only regional geological knowledge.