New Computational Technologies for the Design of 3D Seismic Acquisition Geometries with Compressive Sampling for the Reduction of Economic Costs and Environmental Impacts in Hydrocarbon Exploration in Colombian Onshore Basins
Funding: Agencia Nacional de Hidrocarburos, Ministerio de Ciencia, Tecnología e Innovacion and Fondo Nacional de Financiamiento para la Ciencia, la Tecnología y la Innovación Francisco José de Caldas, Universidad Industrial de Santander, Colombia, Agreement, project No. 110287780575
Background image credit: © Iván Ortiz
Description: This project focused on the development of computational methods based on variational optimization and artificial intelligence, incorporating principles of compressive sampling to optimize resources in terrestrial seismic exploration. Specifically, it aimed to reduce the number of sources required in an orthogonal geometry, where parallel lines of receivers are positioned orthogonally to parallel lines of sources. The intersections of these receiver and source lines generate a 3D seismic datacube in the cross-spread domain.
Personal contribution: I focused on directing and mentoring the research work of two students, as well as ensuring the successful completion of one of the project’s objectives. I guided the development of computational methods based on optimization and deep learning. In one case, I contributed to the development of a seismic shot reconstruction algorithm leveraging non-local self-similarity in seismic slices using a patch-based plug-and-play (PnP) framework with tensor factorization. In the other, we implemented an internal learning neural network to recover missing sources in a 3D seismic datacube, using only the observed sub-sampled data to adaptively learn the optimal parameters.
Associated publications:
Seismic Shot Recovery via Low-Rank Tensor Modeling on the Cross-Spread Domain
31st European Signal Processing Conference (EUSIPCO) (2023).
Ortiz, I., Gelvez-Barrera, T., Galvis, L., Arguello, H.
Seismic Source Recovery Algorithm via Internal Learning in the Cross-Spread Domain
Fourth HGS/EAGE Conference on Latin America, European Association of Geoscientists and Engineers (2022).
Rivera, S., Ortiz, I., Gelvez-Barrera, T., Galvis, L., Arguello, H.