Google and NASA Release AI Model That Detects 50% More Methane Plumes Than Human Analysts

Google and NASA Launch Advanced Methane Detection System

Google Research and NASA's Jet Propulsion Laboratory have released MAPL-EMIT, a deep learning model that maps global methane emissions from space with unprecedented precision1

1

2

2

. Published in PNAS, the AI model detects methane plumes at 60-meter resolution using NASA's EMIT instrument mounted on the International Space Station1

1

. The breakthrough system identifies 50% more plumes than human analysts and has discovered 23,000 additional methane plumes globally that were previously undetected2

2

.

Vision Transformer Trained on Millions of Simulated Plumes

The AI model employs a vision transformer architecture trained on 3.6 million physics-simulated methane plumes1

1

2

2

. This extensive training enables MAPL-EMIT to cut through complex, noisy terrain and read raw spectra from NASA's EMIT instrument at 60 meters per pixel1

1

. The deep learning model significantly outperforms existing NASA products, detecting plumes at 24 of the world's 25 largest-emitting landfills compared to just 17 for the previous system1

1

. UCLA compiled the list of worst emitting sites using Carbon Mapper analysis of EMIT and Planet Labs data1

1

.

Critical Gap in Climate Mitigation Revealed

Methane poses a severe climate threat with warming potential 30 times greater than carbon dioxide over a 100-year timeframe2

2

. MAPL-EMIT tackles a critical bottleneck in environmental monitoring by making methane sources easier to find at scale, enabling faster, more targeted climate mitigation2

2

. The 60-meter resolution allows the system to resolve individual landfill cells, providing facility-level attribution that was previously impossible1

1

. This granular detail matters because it identifies specific emission sources rather than broad regional hotspots.

Europe's Regulatory Framework Excludes Major Emitters

Europe's Methane Regulation mandates a satellite monitoring tool and super-emitter alert system under Article 31, but covers only oil, gas and coal sectors1

1

. Landfills remain outside the regulatory scope despite Prognos and the ifeu Institute estimating that municipal waste landfilled in Europe between 2022 and 2050 will release roughly 1.5 billion tonnes of carbon dioxide equivalent as methane1

1

. The Copernicus atmosphere service operates a methane hotspot explorer built on Sentinel-5P data, but this system covers the planet at roughly 5.5 by 3.5 kilometers per pixel—regional rather than site-specific resolution1

1

. Europe's high-resolution answer remains under development, with AIRMO raising EUR 5M in March to launch its first satellite in early 2027 and the Copernicus CO2M mission scheduled for November 20271

1

.

Open-Source Tools Democratize Access

Google and NASA have made the entire system freely available to researchers, policymakers and operators worldwide2

2

. The global plume database sits on Earth Engine under a Creative Commons license, with an Earth Engine app available to visualize the data1

1

2

2

. Open-source models are accessible on Kaggle and inference tools are published on GitHub1

1

2

2

. This democratized access costs nothing and provides immediate capability for environmental monitoring1

1

. The limitation lies in EMIT's coverage—built to map mineral dust in arid regions, the instrument flies on a space station orbit inclined at 51.6 degrees, meaning Europe was never the primary target1

1

. The sharpest methane map in existence emerged as a by-product of an American dust mission, now given away freely1

1

.