tacofoundation/methaneset

---
license: cc-by-nc-sa-4.0
task_categories:
- image-segmentation
tags:
- earth-observation
- remote-sensing
- methane
- satellite
- taco
pretty_name: MethaneSET
size_categories:
- 100K<n<1M
viewer: false
---

# MethaneSET: Unified Multi-Sensor Datasets for Satellite-Based Methane Plume Detection

**Authors:** Cesar Aybar, Julio Contreras, David Montero, Miguel D. Mahecha, Luis Gómez-Chova

**Paper:** Scientific Data (under review)

Methane is the second-largest driver of anthropogenic warming, and a disproportionate share of emissions comes from a small number of super-emitters detectable by satellite. MethaneSET provides analysis-ready datasets for methane plume detection spanning three sensors with expert-verified segmentation masks from two independent monitoring systems (IMEO MARS and Carbon Mapper). For multispectral sensors (Sentinel-2, Landsat-8/9), labeled scenes with plume masks and confirmed plume-free references are provided for change-detection workflows. For the EMIT imaging spectrometer, calibrated radiance cubes and precomputed matched-filter products support both end-to-end and retrieval-based detection. A synthetic plume bank of over one million WRF-LES enhancements enables physics-based data augmentation. All datasets follow the [TACO specification](https://github.com/tacofoundation/tacoreader) and are distributed as Parquet catalogs with Cloud-Optimized GeoTIFFs.

| Dataset | Samples | Size | Datacard |
|---------|--------:|-----:|----------|
| methaneset-s2-pretraining | 57,291 | 37.2 GB | [view](https://huggingface.co/datasets/tacofoundation/methaneset/blob/main/methaneset-s2-pretraining/index.html) |
| methaneset-s2-finetune | 3,612 | 13.6 GB | [view](https://huggingface.co/datasets/tacofoundation/methaneset/blob/main/methaneset-s2-finetune/index.html) |
| methaneset-l89-pretraining | 21,926 | 8.88 GB | [view](https://huggingface.co/datasets/tacofoundation/methaneset/blob/main/methaneset-l89-pretraining/index.html) |
| methaneset-l89-finetune | 1,548 | 0.78 GB | [view](https://huggingface.co/datasets/tacofoundation/methaneset/blob/main/methaneset-l89-finetune/index.html) |
| methaneset-emit | 503 | 652 GB | [view](https://huggingface.co/datasets/tacofoundation/methaneset/blob/main/methaneset-emit/index.html) |
| methaneset-bank | 1,064,448 | 31.5 GB | [view](https://huggingface.co/datasets/tacofoundation/methaneset/blob/main/methaneset-bank/index.html) |

## Tutorials

| Notebook | Description |
|----------|-------------|
| [](https://colab.research.google.com/github/tacofoundation/methaneset/blob/main/notebooks/01_multispectral.ipynb) | Multispectral plume detection (Sentinel-2 / Landsat-8/9) |
| [](https://colab.research.google.com/github/tacofoundation/methaneset/blob/main/notebooks/02_hyperspectral.ipynb) | Hyperspectral plume detection (EMIT) |

## Citation

```bibtex
@article{aybar2026methaneset,
title = {MethaneSET: Unified Multi-Sensor Datasets for Satellite-Based Methane Plume Detection},
author = {Aybar, Cesar and Contreras, Julio and Montero, David and Mahecha, Miguel D. and G{\'o}mez-Chova, Luis},
journal = {Scientific Data},
year = {2026}
}
```