Analyzing Methane Emissions Using Satellite Imagery and Artificial Intelligence

Methane has a significant impact on climate change as it is a potent greenhouse gas. The energy sector is responsible for 30% of methane emissions worldwide. Effective and continuous monitoring and mitigation of methane emissions are crucial in mitigating the increase of global temperatures. However, current methane emissions monitoring technologies present challenges due to their reliance on the deployment of sensors and personnel near the emission sources, which can be difficult and costly, particularly in the case of large-scale and remote industrial facilities such as those in the energy sector. Methane emissions from these facilities can span vast areas, making traditional monitoring methods time-consuming and resource intensive. Therefore, there is a pressing need for advanced technologies that can provide accurate and real-time methane emissions data from remote and hard-to-reach locations. The dispersion of methane over large sites poses challenges in the emission measurement and quantification. Satellite imagery, such as those provided by the National Aeronautics and Space Administration (NASA) or by the European Space Agency (ESA) with the Copernicus program, can provide a global map of methane concentrations every day. However, interpreting and analyzing such data is still an open challenge. The contribution of this work is multifold. First, it analyzes how the adoption of artificial intelligence and computer vision techniques can support the monitoring and quantification of methane concentrations from satellite imagery, with the aim of supporting mitigation activities for reducing methane emissions. State-of-the-art computer vision and artificial intelligence methods for methane emission quantification using satellite imagery are reviewed. A particular focus is given to methods for enhancing the accuracy of methane quantification using satellite imagery, and for addressing the lack of ground truth data for assessing the quality of the measurement. Second, this paper provides a review of existing data sources for methane emission monitoring and quantification. Satellite imagery datasets from public and private entities are reviewed, as well as tabular datasets that contain processed information from satellite imagery about methane emitters and their geographically referenced information.