Towards Methane Detection Onboard Satellites

The publication titled 'Towards Methane Detection Onboard Satellites' highlights the urgent need for effective methane detection due to its impact on climate change. By employing machine learning techniques onboard satellites, the study proposes a novel method that utilizes unorthorectified data, known as UnorthoDOS, which bypasses traditional preprocessing steps. This innovation not only reduces downlink costs but also maintains performance levels comparable to conventional methods using orthorectified data. The research underscores the importance of rapid methane detection systems in climate mitigation efforts and contributes to the field by releasing model checkpoints and two machine learning-ready datasets derived from the Earth Surface Mineral Dust Source Investigation (EMIT) sensor. This advancement could significantly enhance the capabilities of satellite-based environmental monitoring and response systems.