| dc.description.abstract | The acquisition of highly sparse wavefield data, while maintaining information content, can yield significant gains in acquisition productivity, and enable cost-effective autonomous platforms. To achieve these goals, the use of wavefield data jointly with its first- and second-order spatial derivatives allows for relaxation of Nyquist sampling criteria up to a factor of three under regular sampling. This sampling rate can be potentially further decimated by means of compressive sensing and preconditioning approaches. We introduce a multicomponent reconstruction scheme which uses sparse, irregularly-sampled data, containing pressure fields and two orders of its spatial derivatives, to retrieve the desired regularly-sampled, dense data. Here, we employ a jittered sampling scheme, together with a consistent preconditioning for the solver, that aims for higher or lower sampling according to the expected data complexity. The method is tested on a seismic shot-gather decimated in space domain with different sampling patterns. The results verify that by regular sampling with one-third of the original Nyquist rate, the resulting decimated wavefield from the seismic data set is successfully reconstructed by the multicomponent reconstruction. When using irregular sampling, our approach provides observably better reconstructions for the cases that are over-subsampled, with sampling rates lower than one-third of the original Nyquist rate, and considerably lower for the lesser complex portions of the data. The multicomponent reconstruction is adapted for 3D wavefield generated from a 2D ocean turbulence model with layer wavespeed configuration for imposing additional derivatives in shot direction and cross-derivatives, along with physics-based preconditioning. The reconstruction has major seismic events well-reconstructed from the over-subsampled wavefield. | |
