Near-Surface P Velocity Estimation using 3D Deconvolution Operators

Near surface characterization is a critical challenge for seismic imaging in many onshore areas. This is frequently the case in the Middle East and especially in southern Oman, where, in addition to lateral heterogeneities creating near surface scattering, the near surface is characterized by an alternance of highly contrasted fast and slow velocity layers creating complex carpets of multiples. While shallow Vs models can be assessed by surface wave inversion, estimating the shallow Vp model remains much more challenging. Indeed, first break tomography struggles with velocity inversions, and reflection tomography lacks shallow picks due to the sparseness of acquisition geometries at short offsets. We propose here to use 3D wide-azimuth virtual gathers obtained by a 3D deconvolution algorithm. When applied on modern dense acquisitions, densely sampled 3D virtual gathers free of surface multiples can be obtained. A simplified version based on a local 1D assumption can also be derived and applied to less dense acquisitions. We present here two case studies from southern Oman, where the full and the simplified versions of deconvolution-based 3D virtual gathers are computed and used to estimate a high-resolution shallow reflectivity and a shallow Vp model by high-definition tomography.