Using a deep neural network to detect methane point sources and quantify emissions from PRISMA hyperspectral satellite images

• Published in: Atmospheric measurement techniques 2023-05, Vol.16 (10), p.2627-2640
• Main Authors: Joyce, Peter, Ruiz Villena, Cristina, Huang, Yahui, Webb, Alex, Gloor, Manuel, Wagner, Fabien H, Chipperfield, Martyn P, Barrio Guilló, Rocío, Wilson, Chris, Boesch, Hartmut
• Format: Article
• Language: English
• Subjects: Accuracy; Air quality management; Anthropogenic factors; Artificial neural networks; Climate change; Clustering; Deep learning; Detection; Emissions; Emissions (Pollution); Emissions control; Environmental aspects; Fossils; Fourier transforms; Future climates; Global temperature changes; Global warming; Global warming potential; Greenhouse gases; Human influences; Hyperspectral imaging; Industrial plant emissions; Instruments; Large eddy simulation; Large eddy simulations; Locating; Meteorological research; Methane; Methane emissions; Neural networks; Nitrogen dioxide; Plumes; Point sources; Radiation; Radiation measurement; Resolution; Satellite data; Satellite imagery; Satellite-borne instruments; Satellites; Sensors; Spatial discrimination; Spatial resolution; Synthane; Water pollution; Wavelengths; Weather forecasting
• ISSN: 1867-1381; 1867-8548
• DOI: 10.5194/amt-16-2627-2023

Anthropogenic emissions of methane (CH.sub.4) have made a considerable contribution towards the Earth's changing radiative budget since pre-industrial times. This is because large amounts of methane are emitted from human activities, and the global warming potential of methane is high. The majority of anthropogenic fossil methane emissions to the atmosphere originate from a large number of small (point) sources. Thus, detection and accurate, rapid quantification of such emissions are vital to enable the reduction of emissions to help mitigate future climate change. There exist a number of instruments on satellites that measure radiation at methane-absorbing wavelengths, which have sufficiently high spatial resolution that can be used for detecting plumes of highly spatially localised methane "point sources" (areas on the order of m.sup.2 to km.sup.2). Searching for methane plumes in methane-sensitive satellite images using classical methods, such as thresholding and clustering, can be useful but is time-consuming and often involves empirical decisions. Here, we develop a deep neural network to identify and quantify methane point source emissions from hyperspectral imagery from the PRecursore IperSpettrale della Missione Applicativa (PRISMA) satellite with 30 m spatial resolution. The moderately high spectral and spatial resolution, as well as considerable global coverage and free access to data, makes PRISMA a good candidate for methane plume detection. The neural network was trained with simulated synthetic methane plumes generated with the large eddy simulation extension of the Weather Research and Forecasting model (WRF-LES), which we embedded into PRISMA images. The deep neural network was successful at locating plumes with a F1 score, precision, and recall of 0.95, 0.96, and 0.92, respectively, and was able to quantify emission rates with a mean error of 24 %. The neural network was furthermore able to locate several plumes in real-world images. We have thus demonstrated that our method can be effective in locating and quantifying methane point source emissions in near-real time from 30 m resolution satellite data, which can aid us in mitigating future climate change.