MethaneScan AI

MethaneScan AI is a satellite-based methane leak detection service from Geofinancial Analytics (GEO) for oil and gas producers. It targets super-emitter events (above 100 kg/hr), which GEO says account for about half of fugitive methane emissions, while onsite continuous monitoring typically covers only around 10% of sites, leaving most unmanned sites unobserved between scheduled inspections.

The service scans every customer site weekly using open data from ESA and NASA satellites; GEO's NSF-funded research names Sentinel-2 and the Landsat suite for plume detection and TROPOMI (Sentinel-5P) for baseline observations. A compute-efficient deep learning model detects and quantifies methane plumes at 30 m resolution, with a detection limit of about 150-200 kg/hr, and attributes emissions to geo-located facilities. GEO scientists review each observed plume and rate it for confidence. Coverage is anywhere onshore, subject to cloud cover, with history back to 2015.

It is sold as a one-month trial for up to 125 sites (US$2,400, including platform analytics, cloud computing, expert analysis and reporting) and as Bespoke Analysis packages covering 2015 to the latest week, which add monthly 1 km MethaneScan Heat Maps and one or four days of expert analysis. GEO prices it at one tenth the cost of aerial flyovers and recommends pairing it with onsite monitors at high-value sites. Stated uses include selling low-methane natural gas, prioritising well abandonment by leak risk and validating carbon-credit permanence claims. Development is funded by the U.S. National Science Foundation, including a US$1.25 million SBIR Phase II award in 2026 to add ground monitoring.