Self-Supervised Free-Surface Multiple Suppression

Free-surface multiples pose significant challenges in seismic data processing, often obscuring primary reflections and reducing imaging quality. Traditional methods rely on computationally expensive algorithms and/or accurate wavelet estimation, while supervised learning approaches require clean labels, which are impractical for real data. Thus, we propose a self-supervised learning framework for free-surface multiple suppression, leveraging multi-dimensional convolution to generate multiples from the observed data and a two-stage training strategy comprising a warm-up and an iterative data refinement stage. The framework eliminates the need for labeled data by iteratively refining predictions using so the network learns to remove the multiples-augmented inputs and pseudo-labels. Numerical examples demonstrate that the proposed method effectively suppresses free-surface multiples while preserving primary reflections. Migration results confirm its ability to reduce artifacts and improve imaging quality.