Efficient proxy for time-lapse seismic forward modeling using a U-net encoder-decoder approach

Michael Diniz, Masoud Maleki, Marcos V. M. Cirne, Shahram Danaei, João Oliveira, Denis José Schiozer, Alessandra Davólio, Anderson Rocha

Abstract: Highlights•An efficient encoder–decoder model was proposed for 4D seismic forward modeling.•The proposed proxy surpasses baseline models when applied to the UNISIM IV benchmark.•Methods of explainable AI were applied to interpret the proxy model operation.•Methods were proposed to interpret the results and the correlation between variables.