Surface-related multiple attenuation based on a self-supervised deep neural network with local wavefield characteristics
Bibliographic record
Abstract
ABSTRACT Multiple suppression is a very important step in seismic data processing. To suppress surface-related multiples, we develop a self-supervised deep neural network method based on a local wavefield characteristic loss function (SDNN-LWCLF). The first and second input data and the output data of the self-supervised deep neural network (SDNN) are the predicted surface-related multiples, the full-wavefield data, and the estimated true surface-related multiples, respectively. The role of the SDNN is to replace the convolutional filter part of adaptive subtraction. Although there are differences in amplitudes and phases between the predicted and true surface-related multiples, the predicted surface-related multiples correspond kinematically to the true surface-related multiples and can be mapped to the estimated true surface-related multiples by the SDNN. The SDNN-LWCLF uses a local wavefield characteristic (LWC) loss function with physical properties to constrain the nonlinear optimization process. The LWC loss function is composed of the mean-absolute-error (MAE) and local normalized crosscorrelation (LNCC) loss functions. LNCC can measure the local similarity between the estimated multiples and the estimated primaries. By minimizing the LWC loss function, the MAE loss function corrects amplitudes and phases of the predicted surface-related multiples to their true values, and the LNCC loss function automatically checks and reduces the leaked multiples and residual primaries in the estimated true surface-related multiples. Our SDNN-LWCLF method does not need label data, such as true primaries and true surface-related multiples, which are usually unavailable in real-world applications. Therefore, the SDNN-LWCLF solves the problem of missing training data. Synthetic and field data examples demonstrate that our method can well suppress the surface-related multiples, and its suppression effect is better than the traditional L1-norm adaptive subtraction method and the SDNN method based on only the MAE loss function.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".