Suppressing Near-Surface Reverse Time Migration Artifacts Using a Band-selecting Imaging Condition

In this paper, I introduce a frequency band-selecting imaging condition for Reverse Time Migration (RTM) of shallow reflection datasets which effectively eliminates artifacts that are problematic on near-surface reflection sections. Within the classically defined near-surface region of the subsurface, standard applications of RTM suffer from low-spatial-frequency artifacts, which are obvious for settings where the impedance contrast is sharp or where the shallow structure is complex. The principal reason for this phenomenon is that commonly used RTM imaging conditions---the cross-correlation imaging condition and Poynting vector imaging condition---do not take into account the fact that near-surface datasets commonly have broad frequency bands. The solution to this problem is application of a band-selected imaging condition. The band-selected imaging condition applies a spatially dependent band-pass filter to both the forward-propagating wavefield and the backward-propagating wavefield at each imaging grid point before the cross-correlation of these two wavefields. Since resolution of migrated seismic data is a function of frequency, this imaging condition minimizes the negative impact on resolution most high-resolution migrated data suffers. The utility of this method and improvement in imaging quality is demonstrated with examples of synthetic and real shallow reflection data. Numerical examples demonstrate the notable effect on resolution and artifacts with the band-selection imaging condition compared to the cross-correlation imaging condition and Poynting vector imaging condition. This new method of RTM clearly demonstrates on real data a mechanism for maintaining data resolution while increasing the accuracy of fine structure details interpreted on shallow reflection data.