About the project

MethaneSET is a benchmark of multi-sensor datasets for the detection and quantification of methane leaks from space, built within the project A Foundational Methane Detection Dataset: Transparent Access to Cloud Optimized Spatio-Temporal Datasets (TACO). It unifies Sentinel-2, Landsat-8/9 and EMIT under one TACO schema, with plume annotations from two independent operational systems (IMEO MARS and Carbon Mapper) and a synthetic plume bank derived from WRF-LES simulations.

Seven datasets are publicly available, totaling approximately 1.07 TB: 83,168 multispectral image pairs (4,905 with expert-verified plume masks and 78,263 confirmed plume-free), 721 EMIT granules with calibrated radiance and two independent masks each, and 238,545 synthetic enhancements plus the 1,647 source simulation cubes. The project builds on the TACO specification to make methane data FAIR: findable, accessible, interoperable and reusable.

Grant

Acknowledgements

This project was funded by the Climate Change AI (CCAI) Innovation Grants program, hosted by CCAI with the support of the Global Methane Hub (GMH).
MethaneSET is also supported by the Spanish Ministry of Science, Innovation and Universities (grants PID2023-148485OB-C21 / C22 funded by MCIU / AEI / 10.13039 / 501100011033, ERDF, EU).
We thank UNEP's International Methane Emissions Observatory (IMEO) and Carbon Mapper for making their plume detections publicly available. We thank Mateo-Garcia et al. 2026 for the verified non-detections of Sentinel-2 and Landsat.
EMIT Level 1B radiance data were obtained from NASA's Land Processes Distributed Active Archive Center (LP DAAC).