ISSN 2756-3251
Advanced in Environmental Science and Technology | Vol. 17 (5), pp. 192-204, May, 2026
Research Article
Hyperspectral Satellite Remote Sensing and Machine Learning for Early Detection of Subsurface Methane Leaks Across Natural Gas Transmission Networks
Carlos Mendoza¹*, Priya Subramanian², Henrik Lindqvist³
¹Department of Geomatics Engineering, National Autonomous University of Mexico, Mexico City, Mexico
²Centre for Atmospheric and Oceanic Sciences, Indian Institute of Science, Bengaluru, India
³Department of Physical Geography and Ecosystem Science, Lund University, Lund, Sweden
Corresponding Author: Carlos Mendoza — [email protected]
Abstract
Fugitive methane emissions from aging natural gas pipelines represent a critical, high-potency driver of near-term radiative forcing, yet ground-based leak surveys remain geographically constrained and infrequent. This investigation integrates short-wave infrared (SWIR) hyperspectral satellite imagery with a gradient-boosted spatial anomaly detection algorithm to pinpoint low-rate subsurface pipeline leaks across semi-arid and agricultural corridors. Validated against controlled blind methane release trials, the framework reliably detected emission rates as low as 18 kg/h with a spatial localization accuracy of 12 meters and a false-positive rate below 4.5%. Atmospheric dispersion inversion coupled with high-resolution wind reanalysis enabled automated flux quantification without ground instrumentation. Operational deployment across 4,200 kilometers of transmission infrastructure identified 37 previously unreported super-emitter plumes, demonstrating a scalable verification framework for national greenhouse gas inventories.
Keywords: methane detection; hyperspectral remote sensing; machine learning; fugitive emissions; pipeline monitoring; greenhouse gases
Manuscript Timeline: Received 19 March, 2026; Revised 22 April, 2026; Accepted 02 May, 2026; Published 08 May, 2026.
Citation: Mendoza, C., Subramanian, P., and Lindqvist, H. (2026). Hyperspectral Satellite Remote Sensing and Machine Learning for Early Detection of Subsurface Methane Leaks Across Natural Gas Transmission Networks. Advanced in Environmental Science and Technology, 17(5), pp. 192-204.
Subscribe to read the full article: https://www.internationalscholarsjournals.org/subscribe-to-read