README.md

---
tags:
- seismic
- multiples
- denoising
- marine-seismic
- geophysics
- synthetic
task_categories:
- image-to-image
- other
size_categories:
- 1G-10G
pretty_name: Marine Multiples Attenuation Dataset
viewer: false
---

# Marine Multiples Attenuation Dataset

Paired noisy-input / multiples-noise-label SEG-Y volumes for supervised marine multiples attenuation.

## Task

**Noise-label regression**: given a noisy pre-stack shot gather, predict the additive multiples component. The denoised signal is recovered as:

```text
denoised = noisy_input - predicted_multiples
```

The uploaded noise label is the supervised target. The clean reference used by the benchmark is computed as:

```text
clean_reference = noisy_input - multiples_label
```

## Dataset Description

- **Noisy input**: `noisy/total_nodw.sgy`
- **Multiples label**: `noise/multiples.sgy`
- **Geometry used by benchmark configs**: 638 traces per shot, 1,976 time samples per trace
- **Format**: Pre-stack SEG-Y, paired volumes with matching geometry

### Example 1

<div align="center">

  <p><b>Noisy Data</b></p>
  <img src="assets/multiples/noisy1.png" alt="Example 1 noisy data" width="92%">

  <p><b>Clean Data</b></p>
  <img src="assets/multiples/clean1.png" alt="Example 1 clean data" width="92%">

  <p><b>Multiples Noise Label</b></p>
  <img src="assets/multiples/noise1.png" alt="Example 1 multiples noise label" width="92%">

</div>

<div align="center"><i>representative multiples attenuation sample.</i></div>

### Example 2

<div align="center">

  <p><b>Noisy Data</b></p>
  <img src="assets/multiples/noisy2.png" alt="Example 2 noisy data" width="92%">

  <p><b>Clean Data</b></p>
  <img src="assets/multiples/clean2.png" alt="Example 2 clean data" width="92%">

  <p><b>Multiples Noise Label</b></p>
  <img src="assets/multiples/noise2.png" alt="Example 2 multiples noise label" width="92%">

</div>

<div align="center"><i>representative multiples attenuation sample.</i></div>

### Example 3

<div align="center">

  <p><b>Noisy Data</b></p>
  <img src="assets/multiples/noisy3.png" alt="Example 3 noisy data" width="92%">

  <p><b>Clean Data</b></p>
  <img src="assets/multiples/clean3.png" alt="Example 3 clean data" width="92%">

  <p><b>Multiples Noise Label</b></p>
  <img src="assets/multiples/noise3.png" alt="Example 3 multiples noise label" width="92%">

</div>

<div align="center"><i>representative multiples attenuation sample.</i></div>


## File Structure

| Kind | Path | Size |
|------|------|------|
| noise | `noise/multiples.sgy` | 3161.4 MB |
| noisy | `noisy/total_nodw.sgy` | 3161.4 MB |

**Total**: 1 noisy + 1 noise SEG-Y files

## Loading Data

```python
import segyio
import numpy as np

def read_shot_gather(path, traces_per_shot=638):
    '''Read a regular SEG-Y file into (n_shots, n_traces, n_time).'''
    with segyio.open(path, "r", strict=False) as src:
        n_traces_total = src.tracecount
        n_shots = n_traces_total // traces_per_shot
        n_time = src.samples.size
        data = np.zeros((n_shots, traces_per_shot, n_time), dtype=np.float32)
        for i in range(n_shots):
            for j in range(traces_per_shot):
                data[i, j, :] = src.trace[i * traces_per_shot + j]
    return data

noisy = read_shot_gather("noisy/total_nodw.sgy", traces_per_shot=638)
multiples = read_shot_gather("noise/multiples.sgy", traces_per_shot=638)
clean_reference = noisy - multiples
```

With `huggingface_hub`:

```python
from huggingface_hub import hf_hub_download

noisy_path = hf_hub_download(
    repo_id="GeoBrain/multiples",
    filename="noisy/total_nodw.sgy",
    repo_type="dataset",
)
```

## Benchmark Split

The companion benchmark uses shot-level FFID splitting to avoid trace leakage. The current multiples configs use:

| Split | Shots |
|-------|-------|
| Train | 510 |
| Val | 64 |
| Test | 64 |

The split is done at loading time, so users can adjust it in their own configs.

## Preprocessing Recipe

The companion benchmark applies:

1. **Normalization**: `max_abs`, global scope on the noisy input; the same scale is applied to the multiples label.
2. **Patching**: overlapping 2D patches on the trace-time plane.
3. **Metric handling**: SNR can skip near-zero clean-reference patches with `min_signal_energy`.

No spherical-divergence correction is applied in the denoising training script.

## Citation

If you use this dataset, cite the dataset repository:

```bibtex
@misc{marine_multiples_attenuation,
  title={Marine Multiples Attenuation Dataset},
  howpublished={https://huggingface.co/datasets/GeoBrain/multiples},
}
```

## References

- `segyio` library: https://github.com/equinor/segyio