Methane Plumes Research Articles

Methane distribution above the Emperor Seamount chain

Dissolved methane (CH4) concentrations were measured in the water column at 25 stations along the four sections above Koko and Jingu guyots (the southern part of Emperor Seamount chain). The measured methane concentrations were relatively low (1–6.5 nM). The patterns of CH4 vertical distributions over Koko and Jingu were different. The greatest dissolved methane concentrations (6.5 nM) were found in the near-bottom layer (357 m) above the Koko summit. For Koko guyot, the greatest CH4 content (3.9–6.5 nM) was mainly associated with the subsurface (10–300 m) layer above the summit. However, another methane plume (6 nM) was detected at 1000 m on the western slope of the guyot. We propose that methane maximum was caused by the influence of the Kuroshio Extension or deep eddies. The CH4 distribution over Jingu gyuot was similar to that in open ocean waters. The greatest methane concentrations (3.8–6 nM) were found in the subsurface layers above the summit. Methane exceeded atmospheric equilibrium concentration in the surface (5–52 % supersaturation) layer for both Koko and Jingu. The methane content and supersaturation level in the subsurface layer was at least two times higher than previously measured values for the open ocean. We believe that the high methane and supersaturation level was caused by enhanced methanogenesis in the water column above the seamounts. The estimated methane flux to the atmosphere varied from 1.4 to 16.3 μmol m−2 day−1 for Koko and from 0.5 to 6.5 μmol m−2 day−1 for Jingu, respectively. The average fluxes calculated for Koko (8.37 μmol m−2 day−1) and Jingu (2.8 μmol m−2 day−1) were significantly greater than the average flux for open and coastal oceans. Given the substantial methane atmospheric contributions, the global methane budget should be reconsidered.

Olfactory-based powered descent guidance for Mars methane plume exploration in long-time-average wind environment

This paper investigates an olfactory-based powered descent guidance method that enable a lander to autonomously locate and target the vent source of any methane plume in long-time-average wind environment on Mars. The episodic methane plumes emanating from the Martian surface invite the possibility of direct access to the subsurface methane reservoir to acquire pristine organic materials. Constrained by the insufficient knowledge of the accurate plume vent location, Mars landers must infer the target location — the plume source — autonomously during the powered descent. However, existing powered descent guidance methods require the target location as a terminal constraint, rendering them ineffective in plume exploration missions. We propose an olfactory-based powered descent guidance method that could steer a lander to target the methane vent source by tracking the gradient of methane concentration in long-time-average wind environment, imposing sub-optimal fuel consumption. A novel olfactory navigation method based on tracking the negative logarithmic concentration gradient, and the corresponding gradient measurement method, is proposed. By implementing the negative logarithmic gradient field, gradient-dependent dynamic equations for the powered descending lander are proposed. The gradient-dependent equations provide a way to transform the original optimal guidance problem, which is non-convex and terminal-free, into a convex and terminal-fixed problem. A suboptimal guidance law, similar to E-guidance, is then derived by solving this problem. The proposed guidance could target the vent source of any methane plume in long-time-average wind environment, imposing modest fuel consumption by feeding back the concentration gradient. Numerical simulations illustrate that, in comparison to the terminal-fixed E-guidance and the open-loop fuel-optimal guidance, the lander imposes additional fuel consumption of 2.57% and 12.15% respectively, using the proposed guidance law.

Satellite‐Based Surveys Reveal Substantial Methane Point‐Source Emissions in Major Oil & Gas Basins of North America During 2022–2023

AbstractUtilizing imaging spectroscopy technology to identify methane super‐emitters plays a vital role in mitigating methane emissions in the Oil & Gas (O&G) sector. While earlier research has uncovered significant point‐source methane emissions from O&G production in the US and Canada, which are key regions with large methane emissions, a comprehensive post‐COVID‐19 survey has been notably absent. Here, we perform a detailed survey of methane super‐emitters across multiple basins of North America (Marcellus Shale, Haynesville/Bossier Shale, Permian Basin and Montney Shale) using the new Chinese Gaofen5‐01A/02 (GF5‐01A/02) satellite measurements during 2022–2023. We detect 139 individual methane plumes emanating from 122 point sources, with flux rates ranging from 519 to 16,071 kg hr−1. These emissions exhibit a highly skewed and heavy‐tailed distribution, constituting approximately 23% of the flux inversion with TROPOMI in the sample region, with a range of 13%–40%. Moreover, we observe a 66.7% reduction in methane emissions in Permian Basin during COVID‐19, followed by fluctuations until spring 2023. By summer 2023, methane emissions rebound to twice their previous magnitude (1.68 ± 0.58 Tg a−1). Using these point‐source surveys, we further quantify a regional methane emission of 2.69 ± 0.86 Tg a−1 in Permian Basin. This estimation closely aligns with top‐down inversions (2.22 ± 0.40 Tg a−1) from TROPOMI. The upscale estimation underscores the effectiveness of high‐resolution remote sensing measurements in improving bottom‐up emissions inventories and refining regional methane emission assessments. Our results highlight the potential climate benefits derived from regular monitoring and specific remediation efforts focused on relatively few strong point‐source emissions.

FMAW-YOLOv5s: A deep learning method for detection of methane plumes using optical images

Natural gas hydrates stored in the subsurface seabed of continental margins are one of the most important carbon reservoirs on Earth. Research on natural gas hydrates is of great significance to global warming and ecological protection. Methane plumes caused by crustal dynamics are usually considered as a sign of existence of natural gas hydrates. Detection of methane plumes thus becomes the first step of cold seep research. This paper conducts comprehensive research on detection of methane plumes based on deep learning methods and optical images. First, we proposed a method of creating high quality and balanced datasets for methane plumes detection tasks using open-source videos. We then proposed a FMAW-YOLOv5s method for methane plumes detection. The FMAW-YOLOv5s method improves the traditional YOLOv5s in design of backbone network, neck network and loss function. The FMAW-YOLOv5s method can realize accurate and fast detection of methane plumes with a precision of 96.9% and FPS of 141.7. The lightweight feature of FMAW-YOLOv5s also enables the deployment in edge computing devices such as AUVs and ROVs. This research can not only promote the study of cold seep activities, but also provide meaningful insights for detection of other underwater events such as gas pipelines leakage.

An Independent Evaluation of GHGSat Methane Emissions: Performance Assessment

AbstractAn independent evaluation of methane emissions data from GHGSat, a private company that operates a constellation of small microsatellites flying Fabry‐Perot spectrometers operating at 1.6 µm, was performed. Data from multiple GHGSat commercial satellites, consisting of retrieved methane, diagnostics, and, where detected, plume and emissions information from roughly 250 scenes across Canada were analyzed. From these, 10 scenes contained methane plumes with a 2% detection rate for oil and gas scenes, and 10% for landfills. Methane precision was found to be 5%/2% on average for the C1/C2–C5 designs, with some variability due to scene albedo, terrain roughness, and airmass. Synthetic GHGSat plumes, generated using Lagrangian plume dispersion model and GHGSat characteristics, indicates typical detection limits of 240/180 kg/hr(C1/C2–C5), with a best case of roughly 100 kg/hr. Emissions and their uncertainties calculated using an alternative approach were in broad agreement with GHGSat‐reported emissions. Overall, the performance of the GHGSat C2 design (also used for C3 onward) for favorable‐viewing conditions was found to be largely consistent with company‐advertised performance.

Investigation on methane hydrate formation behaviors in deep-sea methane seepage environments

Deep-sea cold seep is a natural phenomenon where methane leaks from the sea floor. Methane can be partially conversed through physical, chemical, and biological processes, but the excess methane that remains uncaptured escapes into the atmosphere and contributes to the greenhouse effect. Physical carbon sequestration via hydrate formation is a fast and high-energy density type of carbon sequestration; however, it has been largely overlooked. This study investigated hydrate formation in authigenic carbonate rocks through deep-sea geological surveys in cold seeps in the South China Sea. Investigations have shown that authigenic carbonate rocks can sequester carbon by blocking the methane plume from the seepage vents. This perspective provides a new physical carbon sequestration mode of carbonate rocks by observing the artificial collection of methane plumes in deep-sea seepage areas. The mode was simplified into two steps including the formation of a thin layer of planar hydrate film on carbonate rocks and a thick layer of hydrate through the stacking of bubbles under the thin film. Salinity, a key distinction between seawater and freshwater, was further investigated for its impact on the carbon sequestration mechanism. Results showed that saline ions in seawater had little impact on the thickening rate of the planar hydrate film but reduced the lateral growth rate of hydrates on the bubble by 60%. This reduction in the lateral growth rate greatly facilitated the three-dimensional growth, increased the hydrate thickness on the bubble, and improved the carbon sequestration efficiency. Hydrate formation on the exposed carbonate rock can effectively impede methane plume migration into upper water and atmosphere, mitigate the greenhouse effect, and present a promising strategy for carbon sequestration in deep-sea methane seepage areas.

Methane plume detection after the 2022 Nord Stream pipeline explosion in the Baltic Sea

On September 26th, 2022, the detonations at the gas pipelines Nord Stream 1 and 2 resulted in some of the largest non-natural releases of methane known. The distribution of methane in the surrounding seawater and the possible effects were not apparent. To trace the pathways of methane we recorded CH4 concentrations and the isotopic signal (δ13C-CH4) in seawater, and air. A week post-explosion, we detected methane concentrations up to 4 orders of magnitude above the natural Baltic Sea background. The released fossil methane created a distinct plume with δ13C-CH4 ratios differing from natural background values. The strong water stratification preserved the distribution pattern initiated by the explosion, shown by the laterally strong concentration gradient within the plume. Our analysis encompasses three stages of the explosion's impact; the initial sea-air methane release, measurements taken during our research expedition one week later, and a third stage triggered by the shift from summer to winter conditions as an outlook on how winter mixing and microbial activity will influence the plume.

U-Plume: automated algorithm for plume detection and source quantification by satellite point-source imagers

Abstract. Current methods for detecting atmospheric plumes and inferring point-source rates from high-resolution satellite imagery are labor-intensive and not scalable with regard to the growing satellite dataset available for methane point sources. Here, we present a two-step algorithm called U-Plume for automated detection and quantification of point sources from satellite imagery. The first step delivers plume detection and delineation (masking) with a U-Net machine learning architecture for image segmentation. The second step quantifies the point-source rate from the masked plume using wind speed information and either a convolutional neural network (CNN) or a physics-based integrated mass enhancement (IME) method. The algorithm can process 62 images (each measuring 128 pixels × 128 pixels) per second on a single 2.6 GHz Intel Core i7-9750H CPU. We train the algorithm using large-eddy simulations of methane plumes superimposed on noisy and variable methane background scenes from the GHGSat-C1 satellite instrument. We introduce the concept of point-source observability, Ops=Q/(UWΔB), as a single dimensionless number to predict plume detectability and source rate quantification error from an instrument as a function of source rate Q, wind speed U, instrument pixel size W, and instrument-dependent background noise ΔB. We show that Ops can powerfully diagnose the ability of an imaging instrument to observe point sources of a certain magnitude under given conditions. U-Plume successfully detects and masks plumes from sources as small as 100 kg h−1 in GHGSat-C1 images over surfaces with low background noise and successfully handles larger point sources over surfaces with substantial background noise. We find that the IME method for source quantification is unbiased over the full range of source rates, while the CNN method is biased towards the mean of its training range. The total error in source rate quantification is dominated by wind speed at low wind speeds and by the masking algorithm at high wind speeds. A wind speed of 2–4 m s−1 is optimal for detection and quantification of point sources from satellite data.

CH4Net: a deep learning model for monitoring methane super-emitters with Sentinel-2 imagery

Abstract. We present a deep learning model, CH4Net, for automated monitoring of methane super-emitters from Sentinel-2 data. When trained on images of 23 methane super-emitter locations from 2017–2020 and evaluated on images from 2021, this model detects 84 % of methane plumes compared with 24 % of plumes for a state-of-the-art baseline while maintaining a similar false positive rate. We present an in-depth analysis of CH4Net over the complete dataset and at each individual super-emitter site. In addition to the CH4Net model, we compile and make open source a hand-annotated training dataset consisting of 925 methane plume masks as a machine learning baseline to drive further research in this field.

Dispersion and fate of methane emissions from cold seeps on Hikurangi Margin, New Zealand

The influence of cold seep methane on the surrounding benthos is well-documented but the fate of dissolved methane and its impact on water column biogeochemistry remains less understood. To address this, the distribution of dissolved methane was determined around three seeps on the south-east Hikurangi Margin, south-east of New Zealand, by combining data from discrete water column sampling and a towed methane sensor. Integrating this with bottom water current flow data in a dynamic Gerris model determined an annual methane flux of 3 x 105 kg at the main seep. This source was then applied in a Regional Ocean Modelling System (ROMS) simulation to visualize lateral transport of the dissolved methane plume, which dispersed over ∼100 km in bottom water within 1 year. Extrapolation of this approach to four other regional seeps identified a combined plume volume of 3,500 km3 and annual methane emission of 0.4–3.2 x 106 kg CH4 y-1. This suggests a regional methane flux of 1.1–10.9 x 107 kg CH4 y-1 for the entire Hikurangi Margin, which is lower than previous hydroacoustic estimates. Carbon stable isotope values in dissolved methane indicated that lateral mixing was the primary determinant of methane in bottom water, with potential methane oxidation rates orders of magnitude lower than the dilution rate. Calculations indicate that oxidation of the annual total methane emitted from the five seeps would not significantly alter bottom water dissolved carbon dioxide, oxygen or pH; however, superimposition of methane plumes from different seeps, which was evident in the ROMS simulation, may have localized impacts. These findings highlight the value of characterizing methane release from multiple seeps within a hydrodynamic model framework to determine the biogeochemical impact, climate feedbacks and connectivity of cold seeps on continental shelf margins.

Methane Retrieval Algorithms Based on Satellite: A Review

As the second most predominant greenhouse gas, methane-targeted emission mitigation holds the potential to decelerate the pace of global warming. Satellite remote sensing is an important monitoring tool, and we review developments in the satellite detection of methane. This paper provides an overview of the various types of satellites, including the various instrument parameters, and describes the different types of satellite retrieval algorithms. In addition, the currently popular methane point source quantification method is presented. Based on existing research, we delineate the classification of methane remote sensing satellites into two overarching categories: area flux mappers and point source imagers. Area flux mappers primarily concentrate on the assessment of global or large-scale methane concentrations, with a further subclassification into active remote sensing satellites (e.g., MERLIN) and passive remote sensing satellites (e.g., TROPOMI, GOSAT), contingent upon the remote sensing methodology employed. Such satellites are mainly based on physical models and the carbon dioxide proxy method for the retrieval of methane. Point source imagers, in contrast, can detect methane point source plumes using their ultra-high spatial resolution. Subcategories within this classification include multispectral imagers (e.g., Sentinel-2, Landsat-8) and hyperspectral imagers (e.g., PRISMA, GF-5), contingent upon their spectral resolution disparities. Area flux mappers are mostly distinguished by their use of physical algorithms, while point source imagers are dominated by data-driven methods. Furthermore, methane plume emissions can be accurately quantified through the utilization of an integrated mass enhancement model. Finally, a prediction of the future trajectory of methane remote sensing satellites is presented, in consideration of the current landscape. This paper aims to provide basic theoretical support for subsequent scientific research.

Exploiting the Matched Filter to Improve the Detection of Methane Plumes with Sentinel-2 Data

Existing research indicates that detecting near-surface methane point sources using Sentinel-2 satellite imagery can offer crucial data support for mitigating climate change. However, current retrieval methods necessitate the identification of reference images unaffected by methane, which presents certain limitations. This study introduces the use of a matched filter, developing a novel methane detection algorithm for Sentinel-2 imagery. Compared to existing algorithms, this algorithm does not require selecting methane-free images from historical imagery in methane-sensitive bands, but estimates the background spectral information across the entire scene to extract methane gas signals. We tested the algorithm using simulated Sentinel-2 datasets. The results indicated that the newly proposed algorithm effectively reduced artifacts and noise. It was then validated in a known methane emission point source event and a controlled release experiment for its ability to quantify point source emission rates. The average estimated difference between the new algorithm and other algorithms was about 34%. Compared to the actual measured values in the controlled release experiment, the average estimated values ranged from −48% to 42% of the measurements. These estimates had a detection limit ranging from approximately 1.4 to 1.7 t/h and an average error percentage of 19%, with no instances of false positives reported. Finally, in a real case scenario, we demonstrated the algorithm’s ability to precisely locate the source position and identify, as well as quantify, methane point source emissions.

Increased methane emissions from oil and gas following the Soviet Union’s collapse

Global atmospheric methane concentrations rose by 10 to 15 ppb/y in the 1980s before abruptly slowing to 2 to 8 ppb/y in the early 1990s. This period in the 1990s is known as the "methane slowdown" and has been attributed in part to the collapse of the former Soviet Union (USSR) in December 1991, which may have decreased the methane emissions from oil and gas operations. Here, we develop a methane plume detection system based on probabilistic deep learning and human-labeled training data. We use this method to detect methane plumes from Landsat 5 satellite observations over Turkmenistan from 1986 to 2011. We focus on Turkmenistan because economic data suggest it could account for half of the decline in oil and gas emissions from the former USSR. We find an increase in both the frequency of methane plume detections and the magnitude of methane emissions following the collapse of the USSR. We estimate a national loss rate from oil and gas infrastructure in Turkmenistan of more than 10% at times, which suggests the socioeconomic turmoil led to a lack of oversight and widespread infrastructure failure in the oil and gas sector. Our finding of increased oil and gas methane emissions from Turkmenistan following the USSR's collapse casts doubt on the long-standing hypothesis regarding the methane slowdown, begging the question: "what drove the 1992 methane slowdown?"

Exploiting the entire near-infrared spectral range to improve the detection of methane plumes with high-resolution imaging spectrometers

Abstract. Remote sensing emerges as an important tool for the detection of methane plumes emitted by so-called point sources, which are common in the energy sector (e.g., oil and gas extraction and coal mining activities). In particular, satellite imaging spectroscopy missions covering the shortwave infrared part of the solar spectrum are very effective for this application. These instruments sample the methane absorption features at the spectral regions around 1700 and 2300 nm, which enables the retrieval of methane concentration enhancements per pixel. Data-driven retrieval methods, in particular those based on the matched filter concept, are widely used to produce maps of methane concentration enhancements from imaging spectroscopy data. Using these maps enables the detection of plumes and the subsequent identification of active sources. However, retrieval artifacts caused by particular surface components may sometimes appear as false plumes or disturbing elements in the methane maps, which complicates the identification of real plumes. In this work, we use a matched filter that exploits a wide spectral window (1000–2500 nm) instead of the usual 2100–2450 nm window with the aim of reducing the occurrence of retrieval artifacts and background noise. This enables a greater ability to discriminate between surface elements and methane. The improvement in plume detection is evaluated through an analysis derived from both simulated data and real data from areas including active point sources, such as the oil and gas (O&G) industry from San Joaquin Valley (US) and the coal mines from the Shanxi region (China). We use datasets from the Precursore IperSpettrale della Missione Applicativa (PRISMA) and the Environmental Mapping and Analysis Program (EnMAP) satellite imaging spectrometer missions and from the Airborne Visible/Infrared Imaging Spectrometer – Next Generation (AVIRIS-NG) instrument. We find that the interference with atmospheric carbon dioxide and water vapor is generally almost negligible, while co-emission or overlapping of these trace gases with methane plumes leads to a reduction in the retrieved concentration values. Attenuation will also occur in the case of methane emissions situated above surface structures that are associated with retrieval artifacts. The results show that the new approach is an optimal trade-off between the reduction in background noise and retrieval artifacts. This is illustrated by a comprehensive analysis in a PRISMA dataset with 15 identified plumes, where the output mask from an automatic detection algorithm shows an important reduction in the number of clusters not related to CH4 emissions.

New evidence for CH4 enhancement in the upper troposphere associated with the Asian summer monsoon

The Asian summer monsoon (ASM) region is a key region transporting air to the upper troposphere (UT), significantly influencing the distribution and concentration of trace gases, including methane (CH4), an important greenhouse gas. We investigate the seasonal enhancement of CH4 in the UT over the ASM region, utilizing retrievals from the Atmospheric Infrared Sounder (AIRS), model simulations and in-situ measurements. Both the AIRS data and model simulation reveal a substantial enhancement in CH4 concentrations within the active monsoon region of up to 3%, referring to the zonal means, and of up to 6% relative to the pre-monsoon season. Notably, the spatial distribution of the CH4 plume demonstrates a southwestward shift in the AIRS retrievals, in contrast to the model simulations, which predict a broader enhancement, including a significant increase to the east. A cross-comparison with in-situ measurements, including AirCore measurements over the Tibetan Plateau and airline sampling across the ASM anticyclone (ASMA), favors the enhancement represented by model simulation. Remarkable CH4 enhancement over the west Pacific is also evidenced by in-situ data and simulation as a dynamical extension of the ASMA. Our findings underscore the necessity for cautious interpretation of satellite-derived CH4 distributions, and highlight the critical role of in-situ data in anchoring the assimilation of CH4.

Offshore methane detection and quantification from space using sun glint measurements with the GHGSat constellation

Abstract. ​​​​​​​The ability to detect and quantify methane emissions from offshore platforms is of considerable interest in providing actionable feedback to industrial operators. While satellites offer a distinctive advantage for remote sensing of offshore platforms which may otherwise be difficult to reach, offshore measurements of methane from satellite instruments in the shortwave infrared are challenging due to the low levels of diffuse sunlight reflected from water surfaces. Here, we use the GHGSat satellite constellation in a sun glint configuration to detect and quantify methane emissions from offshore targets around the world. We present a variety of examples of offshore methane plumes, including the largest single emission at (84 000 ± 24 000) kg h−1 observed by GHGSat from the Nord Stream 2 pipeline leak in 2022 and the smallest offshore emission measured from space at (180 ± 130) kg h−1 in the Gulf of Mexico. In addition, we provide an overview of the constellation's offshore measurement capabilities. We measure a median column precision of 2.1 % of the background methane column density and estimate a detection limit, from analytical modelling and orbital simulations, that varies between 160 and 600 kg h−1 depending on the latitude and season.

Landfill gas collection efficiency: Categorization of data from existing in-situ measurements

Landfill methane emissions are commonly estimated using cover-type dependent default collection efficiency values, with a first-order decay model or measured gas collection. Current default collection efficiencies used in the United States were predominately derived from 4 studies conducted during or prior to 2007 that relied on flux chambers. Flux chambers are limited by small sample sizes, placement restrictions, and the inability to measure emissions from gas or leachate collection systems. Since 2007, over 14 new studies have been completed using more advanced technologies that allow for direct measurement of methane plumes from most or all of a landfill’s surface. On average, these measurements are 2–3 times greater than emissions predicted by current models and collection efficiency defaults. In lieu of measuring emissions from all landfills, updating collection efficiency defaults can bring modeled emissions into better alignment with measurements. To this end, collection efficiency estimates derived from measured data were categorized into cover types and then adjusted to account for cases where whole plume measurement was an amalgamation of multiple cover types. The resultant adjusted default values were 41% for daily cover, 69% for intermediate cover, and 71% for final cover. Direct measurement of landfill methane emissions is preferrable to account for the full range of variables driving landfill emissions, including collection system design and operation. However, applying these updated defaults back into the landfill emission models eliminates underprediction of landfill emissions for the dataset reviewed, and would provide a more accurate estimate of landfill gas emissions where measurements are unavailable.

Performance and sensitivity of column-wise and pixel-wise methane retrievals for imaging spectrometers

Abstract. Strong methane point source emissions generate large atmospheric concentrations that can be detected and quantified with infrared remote sensing and retrieval algorithms. Two standard and widely used retrieval algorithms for one class of observing platform, imaging spectrometers, include pixel-wise and column-wise approaches. In this study, we assess the performance of both approaches using the airborne imaging spectrometer (Global Airborne Observatory) observations of two extensive controlled-release experiments. We find that the column-wise retrieval algorithm is sensitive to the flight line length and can have a systematic low bias with short flight lines, which is not present in the pixel-wise retrieval algorithm. However, the pixel-wise retrieval is very computationally expensive, and the column-wise retrieval algorithms can produce good results when the flight line length is sufficiently long. Lastly, this study examines the methane plume detection performance of the Global Airborne Observatory with a column-wise retrieval algorithm and finds minimum detection limits of between 9 of 10 kg h−1 and 90 % probability of detection between 10 and 45 kg h−1. These results present a framework of rules for guiding proper concentration retrieval selection given conditions at the time of observation in order to ensure robust detection and quantification.

Geostationary satellite observations of extreme and transient methane emissions from oil and gas infrastructure.

We demonstrate geostationary satellite monitoring of large transient methane point sources with the US Geostationary Operational Environmental Satellites (GOES). GOES provides continuous 5- to 10-min coverage of the Americas at 1 to 2 km nadir pixel resolution in two shortwave infrared spectral bands from which large methane plumes can be retrieved. We track the full evolution of an extreme methane release from the El Encino-La Laguna natural gas pipeline in Durango, Mexico on 12 May 2019. The release lasted 3 h at a variable rate of 260 to 550 metric tons of methane per hour and totaled 1,130 to 1,380 metric tons. We report several other detections of transient point sources from oil/gas infrastructure, from which we infer a detection limit of 10 to 100 t h-1. Our results show that extreme releases of methane can last less than an hour, as from deliberate venting, and would thus be difficult to identify and quantify with low-Earth orbit satellites.

Semantic segmentation of methane plumes with hyperspectral machine learning models

Methane is the second most important greenhouse gas contributor to climate change; at the same time its reduction has been denoted as one of the fastest pathways to preventing temperature growth due to its short atmospheric lifetime. In particular, the mitigation of active point-sources associated with the fossil fuel industry has a strong and cost-effective mitigation potential. Detection of methane plumes in remote sensing data is possible, but the existing approaches exhibit high false positive rates and need manual intervention. Machine learning research in this area is limited due to the lack of large real-world annotated datasets. In this work, we are publicly releasing a machine learning ready dataset with manually refined annotation of methane plumes. We present labelled hyperspectral data from the AVIRIS-NG sensor and provide simulated multispectral WorldView-3 views of the same data to allow for model benchmarking across hyperspectral and multispectral sensors. We propose sensor agnostic machine learning architectures, using classical methane enhancement products as input features. Our HyperSTARCOP model outperforms strong matched filter baseline by over 25% in F1 score, while reducing its false positive rate per classified tile by over 41.83%. Additionally, we demonstrate zero-shot generalisation of our trained model on data from the EMIT hyperspectral instrument, despite the differences in the spectral and spatial resolution between the two sensors: in an annotated subset of EMIT images HyperSTARCOP achieves a 40% gain in F1 score over the baseline.