Methane is a greenhouse gas 30 times more potent at trapping heat than carbon dioxide over 100 years. It’s responsible for around 30% of the global rise in temperatures to date. Over half of global methane emissions comes from human activities in three sectors: fossil fuels (mainly oil and gas), waste, and agriculture.
Stopping methane leaks would be a huge boon to our fight against climate change. Thankfully, satellites that can detect methane have recently been put into orbit. But sifting through the data and finding the culprits is still pretty difficult.
Researchers at the University of Oxford have now developed a solution. They’ve created a new tool that automatically detects methane plumes on Earth from orbit by using machine learning with hyperspectral data. This could help identify the main sources of methane around the world and also enable more effective action to reduce greenhouse gas emissions.
Model predictions of methane leaks using data from the AVIRIS aerial mission flown above the Four Corners area in the USA in 2019. Image credit: Open source AVIRIS data (NASA) processed by Vít Růžička.
Acting fast on methane emissions reduction can have an immediate impact on slowing global warming. However, until now, there have been only a few methods to readily map methane plumes from aerial imagery and the processing step is very time-consuming. This is because methane is transparent to the human eye and most satellite sensors.
The new tool by Oxford researchers, together with Trillium Technologies’ NIO.space, an AI company, overcomes these problems by detecting methane plumes in data from hyperspectral satellites. These detect narrower bands than more common multispectral satellites, making it easier to detect the specific signature of methane.
Nevertheless, the sheer volume of data generated posed a significant challenge for processing without the assistance of AI. The researchers trained the model using 167,825 hyperspectral tiles (each representing an area of 1.64 km2) captured by NASA’s aerial sensor AVIRIS over the Four Corners area in the United States
The algorithm was subsequently used on data obtained from additional hyperspectral sensors in orbit, including information gathered by NASA’s latest hyperspectral sensor, EMIT (Earth Surface Mineral Dust Source Investigation mission). EMIT is affixed to the International Space Station, offering comprehensive coverage of Earth’s surface.
Earlier this year, using data collected by EMIT, NASA researchers identified more than 50 “super-emitters” of methane in Central Asia, the Middle East, and the Southwestern United States. Super-emitters are facilities, equipment, and other infrastructure, typically in the fossil fuel, waste, or agriculture sectors, that emit a lot of methane.
“What makes this research particularly exciting and relevant is the fact that many more hyperspectral satellites are due to be deployed in the coming years, including from ESA, NASA, and the private sector,” Vít Růžička, lead researcher, said in a press release. “In combination, these new sensors will provide global hyperspectral coverage.”
Example of NASA satellite spotting methane emissions. Image credits: NASA.
Increased accuracy
Overall, the model had an accuracy of over 81% in identifying large methane plumes, surpassing the previous most accurate approach by 21.5%. Also, the method had a substantially improved false positive detection rate for tile classification, reducing it by approximately 41.83% compared to the previously most accurate approach.
In order to encourage continued exploration in methane detection, the researchers have made both the dataset and the model’s code openly available on GitHub. They are currently investigating the feasibility of implementing the model directly onboard the satellite itself, enabling other satellites to perform subsequent observations.
Onboard processing could mean that only priority alerts would need to be sent back to Earth, for instance, a text alert signal with the coordinates of an identified methane source, Vít Růžička said. This would also allow for satellites to collaborate autonomously, with a weak detection from one satellite serving as a tip-off for others.
