Retrieving detailed velocity information from common-offset GPR and CMP-stacked seismic reflection data

We review a velocity estimation technique recently proposed for surface-based common-offset ground-penetrating radar (GPR) reflection data, and we explore its applicability to stacked common-midpoint (CMP) seismic reflection data. The technique seeks to infer the large-scale component of the velocity structure through analysis of the diffracted part of the wavefield, whereas the small-scale component is estimated through inversion of the reflected part. For the latter, a 1D convolutional forward model is used, which, despite its simplicity, is found to be adequate for common-offset GPR reflection data from complex environments and, thus, makes the technique computationally efficient and amenable to large data volumes. Tests on synthetic data for heterogeneous elastic subsurface models demonstrate that the technique is also applicable to CMP-stacked seismic reflection data. This finding is corroborated through an application to the Western Approaches Margin (WAM) marine deep seismic reflection survey acquired offshore southwestern Britain.