Crossline interpolation with the traces-to-trace approach using machine learning

• DOI: 10.1190/segam2020-3428348.1
• Publication Date: 2020-10-01T01:28:46Z
• Authors: Zeu Yeeh; Joongmoo Byun; Daeung Yoon
• Keywords: Interpolation; TRACE (psycholinguistics)

Trace interpolation using machine learning (ML) has been actively studied recently. Especially, crossline in towed streamer system is an important task due to the sparsity of data compared dense inline data. The key successful ML application trace how similar training are target data, which be interpolated. Considering similarity, we use for model, and then apply trained model interpolation. In this way, can train fill gaps on sparse same seismic without additional datasets. For based interpolation, traces-to-trace approach with LSTM (Long Short-term Memory) networks, uses two traces as input one between output. addition, design multiple networks predict at different locations ratios traces. A synthetic used demonstrate effectiveness proposed method networks. Presentation Date: Tuesday, October 13, 2020 Session Start Time: 1:50 PM Location: Poster Station 1 Type: