"Field Seismic Data For Denoising"

"This dataset is established for the training and quantitative evaluation of deep learning-based seismic data denoising methods. It comprises independent pure noise data and authentic field seismic datasets acquired from actual geological exploration sites. By adopting a random superposition scheme, we can extract relatively clean effective seismic components from real field records, and then inject simulated interference with different noise intensities and types into these clean signals. This data construction method can effectively expand the number of training samples, simulate the complex noise distribution characteristics of actual seismic exploration, and provide reliable data support for verifying the robustness and generalization performance of seismic denoising networks."