Automatic Detection Of Methane Emissions In Multispectral Satellite Imagery Using A Vision Transformer

This article discusses the development of a deep learning architecture for detecting methane emissions using multi-spectral satellite data. The results show significant improvement in detection capabilities, with the ability to detect emissions as small as 0.01 km2. The approach also allows for automated monitoring of persistent methane emissions on a global scale. The use of synthetic methane plumes for training data is also discussed.