Published Research

Observations of methane plumes from point sources provide valuable information on emissions and support effective mitigation strategies. This work examines the likelihood of methane plumes being detected by different instruments under different conditions, such as the sensor type, weather, and the emission rate of the source. Results highlight the importance of an accurate probability of detection model for interpretation of point source emissions.

Using observations from MethaneAIR from regions responsible for 70% of 2023 U.S. onshore oil and gas production, this study estimates total oil/gas methane emissions to be ~8 Tg/yr—around five times higher than U.S. EPA estimates and equivalent to a 1.6% methane loss rate. Performance varies widely by basin, with highly productive regions like the Permian, Appalachian, and Haynesville-Bossier showing the largest total emissions, while older basins such as the Uinta and Piceance exhibit higher loss rates. Regional comparisons showed good agreement across total emissions quantified by MethaneAIR and other empirical and remote sensing estimates.
 

Introduces the Spectral Channel Attention Network (SCAN), a deep learning approach designed to improve cloud and cloud shadow detection in hyperspectral imagery from MethaneSAT and MethaneAIR. By dynamically weighting individual spectral bands based on their physical relevance, rather than treating all wavelengths equally, SCAN outperforms traditional U-Net and transformer-based attention models on MethaneSAT data, improving F1-scores and shadow detection accuracy. When combined with spatial models in an ensemble framework, the approach achieves state-of-the-art performance, directly strengthening the reliability of satellite-based methane retrievals worldwide.

Note: This is a preprint, and we expect the final published version to become available soon.

This work demonstrates the capability of MethaneAIR data to be used in tandem with TROPOMI satellite data for inferring methane emissions from an oil and gas basin. The work combines the two instruments using the common platform of the Integrated Methane Inversion (IMI) which is an open source software tool for using computer models of the atmosphere (GOES-Chem) with methane concentration observations to improve knowledge of emissions. We show that estimates of emissions can be improved by including data from both sources.

This study created the first automated artificial intelligence system to find natural gas compressor stations in satellite images. When tested in a large U.S. oil and gas region, the system identified over 1,100 previously unreported facilities, suggesting that existing public databases are missing many sources of pollution. As a result, public exposure to harmful emissions may be underestimated by up to 74%. Results show how machine learning can improve oil and gas infrastructure mapping for tracking and managing methane pollution.

This study created the first automated artificial intelligence system to find natural gas compressor stations in satellite images. When tested in a large U.S. oil and gas region, the system identified over 1,100 previously unreported facilities, suggesting that existing public databases are missing many sources of pollution. As a result, public exposure to harmful emissions may be underestimated by up to 74%. Results show how machine learning can improve oil and gas infrastructure mapping for tracking and managing methane pollution.

A comprehensive assessment of over 400 point source methane emitters detected with MethaneAIR from 2021-2023 across 13 major oil and gas basins covering ~80% of US onshore production was presented in this paper. This was the most geographically extensive survey by an airborne methane imaging spectrometer in a single year and contributes analyses from multiple regions that had not previously been represented in the methane point source literature. Authors describe automated plume-finding methods, perform detailed attribution to facility categories within oil and gas and non-oil and gas sectors, and quantify total point source methane emissions from these basins of 360 t h-1 in 2023, with ~80 % of the total attributable to oil and gas sources.

Presents and validates the retrieval algorithm used by MethaneSAT and MethaneAIR, using MethaneAIR observations over major U.S. oil and gas basins in 2019 and 2021. Repeated surveys reveal both persistent and intermittent high-emitting sources, including a large processing facility with unusually high leak rates and a ruptured pipeline, demonstrating the capability and value of MethaneSAT-style observations for detecting, and quantifying basin-wide methane emissions.

This study presents a sensitive and computationally inexpensive method for detecting methane plumes in MethaneAIR data using a matched-filter algorithm. The performance of the method was demonstrated through comparison with controlled release experiments, comparison with simulated plumes, and intercomparison with other methods. Authors applied this processing chain to MethaneAIR data mosaics acquired over the Permian Basin during flights in 2021 and 2023, which resulted in the detection of hundreds of point sources above 100–200 kg h−1, with a conservative detection limit of around 120 kg h−1.