Recovering low frequencies for impedance inversion by frequency domain deconvolution

Sina Esmaeili, Gary F. Margrave

Acoustic impedance is a rock property that can be derived from seismic data and contains important information about subsurface properties. Direct measurements of acoustic impedance are available from acoustic and density well logs, but these well data can provide the acoustic impedance only at the well's location. Mathematically it is true that acoustic impedance can be calculated from earth's reflectivity function, and this function can be estimated from seismic data. Additionally, estimation of reflectivity from seismic data is always bandlimited and affects acoustic impedance significantly. Acoustic impedance inversion can easily be computed by a standard impedance inversion algorithm which uses well logs to fill in the low-frequency information that is missing in bandlimited seismic data.

In this study we investigate the performance of standard deconvolution and its ability to recover low frequency content directly from seismic data. We find that standard deconvolution does not perform well at low frequencies and this is a limiting factor in impedance inversion. Using frequency domain deconvolution, we show that improving the spectral smoothing process and applying a minimum phase spectral color operator to the deconvolved seismic trace can improve the performance of impedance inversion and reduce the bandwidth necessary from well control.