A Physics-Based Band-Ratio Algorithm for Methane Detection with PRISMA Satellite Data Liew S.C., Tan L., Salinas S.V.
Centre for Remote Imaging, Sensing and Processing (CRISP),
National University of Singapore
Abstract
Greenhouse gases, particularly methane, are primary drivers of climate change, with methane possessing a global warming potential 83 times higher than carbon dioxide over two decades. Consequently, monitoring the spatial and temporal distribution of methane emission from both natural and human sources is crucial for climate modeling, prediction, and validating carbon reduction targets. Satellite sensors offer significant advantages over ground-based methods for this purpose, providing rapid, wide-area coverage, access to remote locations, global monitoring capabilities, high-resolution localized detection, and the ability to track emission trends over time. Satellites detect methane through two primary modes: measuring the absorption of sunlight by methane molecules using hyperspectral imaging in the short-wave infrared (SWIR) bands, or by analyzing natural thermal emission from Earth surface in the mid-wave infrared (MWIR) band. While instruments like TROPOMI on Sentinel-5P offer global observations of methane emission, their spatial resolution is insufficient for localized sources. In a previous work presented in IGARSS 2024, we introduced a physics-based inverse modeling algorithm using the linear matrix inversion technique for retrieving methane concentration of a near-ground methane cloud. In this paper we show that the physics-based algorithm may be simplified to a simple band-ratio method using appropriately selected SWIR spectral bands, for detecting and quantifying methane emissions from hyperspectral PRISMA satellite data. This method was validated using a synthetic dataset and successfully applied to a super-emitter site in Turkmenistan to map methane columnar density. This work demonstrates the importance of physics-based modeling in providing insights for retrieving quantitative information from a simple band-ratio algorithm.
