Search

Abstract. Sentinel-2 satellite imagery has been shown by studies to becapable of detecting and quantifying methane emissions from oil and gasproduction. However, current methods lack performance calibration withground-truth testing. This study developed a multi-band–multi-pass–multi-comparison-date methane retrieval algorithm that enhances Sentinel-2 sensitivity to methane plumes. The method was calibratedusing data from a large-scale controlled-release test in Ehrenberg, Arizona,in fall 2021, with three algorithm parameters tuned based on the trueemission rates. Tuned parameters are the pixel-level concentration upper-bound threshold during extreme value removal, the number of comparisondates, and the pixel-level methane concentration percentage threshold whendetermining the spatial extent of a plume. We found that a low value of theupper-bound threshold during extreme value removal can result in falsenegatives. A high number of comparison dates helps enhance the algorithmsensitivity to the plumes in the target date, but values in excess of12 d are neither necessary nor computationally efficient. A high percentagethreshold when determining the spatial extent of a plume helps enhance thequantification accuracy, but it may harm the yes/no detection accuracy. Wefound that there is a trade-off between quantification accuracy anddetection accuracy. In a scenario with the highest quantification accuracy,we achieved the lowest quantification error and had zero false-positivedetections; however, the algorithm missed three true plumes, which reduced theyes/no detection accuracy. In contrast, all of the true plumes weredetected in the highest detection accuracy scenario, but the emission ratequantification had higher errors. We illustrated a two-step method thatupdates the emission rate estimates in an interim step, which improvesquantification accuracy while keeping high yes/no detection accuracy. Wealso validated the algorithm's ability to detect true positives and truenegatives in two application studies.

We show that the Landsat and Sentinel-2 satellites can detect NO2 plumes from large point sources at 10 to 60 m pixel resolution in their blue and ultrablue bands. We use the resulting NO2 plume imagery to quantify nitrogen oxides (NOx) emission rates for several power plants in Saudi Arabia and the United States, including a 13-y analysis of 132 Landsat plumes from Riyadh power plant 9 from 2009 through 2021. NO2 in the plumes initially increases with distance from the source, likely reflecting recovery from ozone titration. The fine pixel resolutions of Landsat and Sentinel-2 enable separation of individual point sources and stacks, including in urban background, and the long records enable examination of multidecadal emission trends. Our inferred NOx emission rates are consistent with previous estimates to within a precision of about 30%. Sources down to ~500 kg h-1 can be detected over bright, quasi-homogeneous surfaces. The 2009 to 2021 data for Riyadh power plant 9 show a strong summer peak in emissions, consistent with increased power demand for air conditioning, and a marginal slow decrease following the introduction of Saudi Arabia's Ambient Air Standard 2012. [ABSTRACT FROM AUTHOR]

Venezuela has long been identified as an area with large methane emissions and intensive oil exploitation, especially in the Lake Maracaibo region, but production has strongly decreased in recent years. The area is notoriously difficult to observe from space due to its complex topography and persistent cloud cover. We use the unprecedented coverage of the TROPOspheric Monitoring Instrument (TROPOMI) methane observations in analytical inversions with the Integrated Methane Inversion (IMI) framework at the national scale and at the local scale with the Weather Research and Forecasting model with chemistry (WRF-Chem). In the IMI analysis, we find Venezuelan emissions of 7.5 (5.7–9.3) Tga-1 in 2019, where about half of emissions can be informed by TROPOMI observations, and emissions from oil exploitation are a factor of ∼ 1.6 higher than in bottom-up inventories. Using WRF, we find emissions of 1.2 (1.0–1.5) Tga-1 from the Lake Maracaibo area in 2019, close to bottom-up estimates. Our WRF estimate is ∼ 40 % lower than the result over the same region from the IMI due to differences in the meteorology used by the two models. We find only a small, non-significant trend in emissions between 2018 and 2020 around the lake, implying the area's methane emission intensity expressed against oil and gas production has doubled over the time period, to ∼ 20 %. This value is much higher than what has previously been found for other oil and gas production regions and indicates that there could be large emissions from abandoned infrastructure. [ABSTRACT FROM AUTHOR]

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 kgh-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 ms-1 is optimal for detection and quantification of point sources from satellite data. [ABSTRACT FROM AUTHOR]

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?". [ABSTRACT FROM AUTHOR]

Venezuela has long been identified as an area with large methane emissions, especially in the Lake Maracaibo region with intensive oil exploitation, but production has strongly decreased in recent years. The area is notoriously difficult to observe from space due to its complex topography and persistent cloud cover. We use the unprecedented coverage of the TROPOspheric Monitoring Instrument (TROPOMI) methane observations in analytical inversions at the national scale with the Integrated Methane Inversion (IMI) framework and at regional scale with the Weather Research and Forecasting (WRF) model. In the IMI analysis, we find 2019 Venezuelan emissions of 7.5 (5.7–9.3) Tg a-1, where about half of emissions can be informed by TROPOMI observations and emissions from oil exploitation are a factor ~1.6 higher than in bottom-up inventories. Using WRF, we find 2019 emissions of 1.2 (1.0–1.5) Tg a-1 from the Lake Maracaibo area, close to bottom-up estimates. Our WRF estimate is ~40 % lower than the regional result from the IMI due to differences in the meteorology used by the two models. We only find a small, non-significant, trend in emissions between 2018 and 2020 around the lake, implying the area's methane emission intensity expressed against oil/gas production has doubled over the time period to ~20 %. This value is much higher than what has previously been found for other oil/gas production regions and indicates there could be large emissions from abandoned infrastructure. [ABSTRACT FROM AUTHOR]

Current methods for detecting atmospheric plumes and inferring point source rates from high-resolution satellite imagery are labor intensive and not scalable 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 machine learning U-Net architecture for image segmentation. The second step quantifies point source rate from the masked plume using wind speed information and either a convolution neural network (CNN) or a physics-based Integrated Mass Enhancement (IME) method. The algorithm can process 62 128×128 images per second on a single core. We train the algorithm with 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/(UW?B) 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 κB. 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 over surfaces with low background noise and succeeds for 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 toward 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. [ABSTRACT FROM AUTHOR]

Satellite observations of dry-column methane mixing ratios (XCH 4) from shortwave infrared (SWIR) solar backscatter radiation provide a powerful resource to quantify methane emissions in service of climate action. The TROPOspheric Monitoring Instrument (TROPOMI), launched in October 2017, provides global daily coverage at a 5.5 × 7 km 2 (nadir) pixel resolution, but its methane retrievals can suffer from biases associated with SWIR surface albedo, scattering from aerosols and cirrus clouds, and across-track variability (striping). The Greenhouse gases Observing SATellite (GOSAT) instrument, launched in 2009, has better spectral characteristics and its methane retrieval is much less subject to biases, but its data density is 250 times sparser than TROPOMI. Here, we present a blended TROPOMI + GOSAT methane product obtained by training a machine learning (ML) model to predict the difference between TROPOMI and GOSAT co-located measurements, using only predictor variables included in the TROPOMI retrieval, and then applying the correction to the complete TROPOMI record from April 2018 to present. We find that the largest corrections are associated with coarse aerosol particles, high SWIR surface albedo, and across-track pixel index. Our blended product corrects a systematic difference between TROPOMI and GOSAT over water, and it features corrections exceeding 10 ppb over arid land, persistently cloudy regions, and high northern latitudes. It reduces the TROPOMI spatially variable bias over land (referenced to GOSAT data) from 14.3 to 10.4 ppb at a 0.25 ∘ × 0.3125 ∘ resolution. Validation with Total Carbon Column Observing Network (TCCON) ground-based column measurements shows reductions in variable bias compared with the original TROPOMI data from 4.7 to 4.4 ppb and in single-retrieval precision from 14.5 to 11.9 ppb. TCCON data are all in locations with a SWIR surface albedo below 0.4 (where TROPOMI biases tend to be relatively low), but they confirm the dependence of TROPOMI biases on SWIR surface albedo and coarse aerosol particles, as well as the reduction of these biases in the blended product. Fine-scale inspection of the Arabian Peninsula shows that a number of hotspots in the original TROPOMI data are removed as artifacts in the blended product. The blended product also corrects striping and aerosol/cloud biases in single-orbit TROPOMI data, enabling better detection and quantification of ultra-emitters. Residual coastal biases can be removed by applying additional filters. The ML method presented here can be applied more generally to validate and correct data from any new satellite instrument by reference to a more established instrument. [ABSTRACT FROM AUTHOR]

We present a versatile, powerful, and user-friendly chemical data assimilation toolkit for simultaneously optimizing emissions and concentrations of chemical species based on atmospheric observations from satellites or suborbital platforms. The CHemistry and Emissions REanalysis Interface with Observations (CHEEREIO) exploits the GEOS-Chem chemical transport model and a localized ensemble transform Kalman filter algorithm (LETKF) to determine the Bayesian optimal (posterior) emissions and/or concentrations of a set of species based on observations and prior information using an easy-to-modify configuration file with minimal changes to the GEOS-Chem or LETKF code base. The LETKF algorithm readily allows for nonlinear chemistry and produces flow-dependent posterior error covariances from the ensemble simulation spread. The object-oriented Python-based design of CHEEREIO allows users to easily add new observation operators such as for satellites. CHEEREIO takes advantage of the Harmonized Emissions Component (HEMCO) modular structure of input data management in GEOS-Chem to update emissions from the assimilation process independently from the GEOS-Chem code. It can seamlessly support GEOS-Chem version updates and is adaptable to other chemical transport models with similar modular input data structure. A post-processing suite combines ensemble output into consolidated NetCDF files and supports a wide variety of diagnostic data and visualizations. We demonstrate CHEEREIO's capabilities with an out-of-the-box application, assimilating global methane emissions and concentrations at weekly temporal resolution and 2 ∘ × 2.5 ∘ spatial resolution for 2019 using TROPOspheric Monitoring Instrument (TROPOMI) satellite observations. CHEEREIO achieves a 50-fold improvement in computational performance compared to the equivalent analytical inversion of TROPOMI observations. [ABSTRACT FROM AUTHOR]

We quantify weekly methane emissions at 0.25 ∘ × 0.3125 ∘ (≈25 × 25 km 2) resolution from the Permian Basin, the largest oil production basin in the US, by inverse analysis of satellite observations from the TROPOspheric Monitoring Instrument (TROPOMI) from May 2018 to October 2020. The mean oil and gas emission from the region (± standard deviation of weekly estimates) was 3.7 ± 0.9 Tg a -1 , higher than previous TROPOMI inversion estimates that may have used biased prior emissions or background assumptions. We find strong week-to-week variability in emissions superimposed on longer-term trends, and these are consistent with independent inferences of temporal emission variability from tower, aircraft, and multispectral satellite data. New well development and natural gas spot price were significant drivers of variability in emissions over our study period but the concurrent 50 % increase in oil and gas production was not. The methane intensity (methane emitted per unit of methane gas produced) averaged 4.6 % ± 1.3 % and steadily decreased from 5 %–6 % in 2018 to 3 %–4 % in 2020. While the decreasing trend suggests improvement in operator practices during the study period, methane emissions from the Permian Basin remained high, with methane intensity an order of magnitude above the industry target of <0.2 %. Our success in using TROPOMI satellite observations for weekly estimates of emissions from a major oil production basin shows promise for application to near-real-time monitoring in support of climate change mitigation efforts. [ABSTRACT FROM AUTHOR]

The United States is the world's largest oil/gas methane emitter according to current national reports. Reducing these emissions is a top priority in the US government's climate action plan. Here, we use a 2010 to 2019 high-resolution inversion of surface and satellite observations of atmospheric methane to quantify emission trends for individual oil/gas production regions in North America and relate them to production and infrastructure. We estimate a mean US oil/gas methane emission of 14.8 (12.4 to 16.5) Tg a-1 for 2010 to 2019, 70% higher than reported by the US Environmental Protection Agency. While emissions in Canada and Mexico decreased over the period, US emissions increased from 2010 to 2014, decreased until 2017, and rose again afterward. Increases were driven by the largest production regions (Permian, Anadarko, Marcellus), while emissions in the smaller production regions generally decreased. Much of the year-to-year emission variability can be explained by oil/gas production rates, active well counts, and new wells drilled, with the 2014 to 2017 decrease driven by reduction in new wells and the 2017 to 2019 surge driven by upswing of production. We find a steady decrease in the oil/gas methane intensity (emission per unit methane gas production) for almost all major US production regions. The mean US methane intensity decreased from 3.7% in 2010 to 2.5% in 2019. If the methane intensity for the oil/gas supply chain continues to decrease at this pace, we may expect a 32% decrease in US oil/gas emissions by 2030 despite projected increases in production. [ABSTRACT FROM AUTHOR]

We present a versatile, powerful, and user-friendly chemical data assimilation toolkit for simultaneously optimizing emissions and concentrations of chemical species based on atmospheric observations from satellites or suborbital platforms. The CHemistry and Emissions REanalysis Interface with Observations (CHEEREIO) exploits the GEOS-Chem chemical transport model and a localized ensemble transform Kalman filter algorithm (LETKF) to determine the Bayesian optimal (posterior) emissions and/or concentrations of a set of species based on observations and prior information, using an easy-to-modify configuration file with minimal changes to the GEOS-Chem or LETKF code base. The LETKF algorithm readily allows for non-linear chemistry and produces flow-dependent posterior error covariances from the ensemble simulation spread. The object-oriented Python-based design of CHEEREIO allows users to easily add new observation operators such as for satellites. CHEEREIO takes advantage of the HEMCO modular structure of input data management in GEOS-Chem to update emissions from the assimilation process independently from the GEOS-Chem code. It can seamlessly support GEOS-Chem version updates and is adaptable to other chemical transport models with similar modular input data structure. A postprocessing suite combines ensemble output into consolidated NetCDF files and supports a wide variety of diagnostic data and visualizations. We demonstrate CHEEREIO's capabilities with an out-of-the-box application, assimilating global methane emissions and concentrations at weekly temporal resolution and 2°x2.5° spatial resolution for 2019 using TROPOMI satellite observations. CHEEREIO achieves a 50-fold improvement in computational performance compared to the equivalent analytical inversion of TROPOMI observations. [ABSTRACT FROM AUTHOR]

We use 2019 TROPOMI satellite observations of atmospheric methane in an analytical inversion to quantify methane emissions from the Middle East and North Africa at up to ~25 km × 25 km resolution, using spatially allocated national UNFCCC reports as prior estimates for the fuel sector. Our resulting best estimate of anthropogenic emissions for the region is 35 % higher than the prior bottom-up estimate (+103 % for gas, +53 % for waste, +49 % for livestock, −14 % for oil) with large variability across countries. Oil and gas account for 38 % of total anthropogenic emissions in the region. TROPOMI observations can effectively optimize and separate national emissions by sector for most of the 23 countries in the region, with 6 countries accounting for most of total anthropogenic emissions including Iran (5.3 (5.0–5.5) Tg a−1; best estimate and uncertainty range), Turkmenistan (4.4 (2.8–5.1) Tg a−1), Saudi Arabia (4.3 (2.4–6.0) Tg a−1), Algeria (3.5 (2.4–4.4) Tg a−1), Egypt (3.4 (2.5–4.0) Tg a−1) , and Turkey (3.0 (2.0–4.1) Tg a−1). Most oil/gas emissions are from the production (upstream) subsector, but Iran, Turkmenistan, and Saudi Arabia have large gas emissions from transmission and distribution subsectors. We identify a high number of annual oil/gas emission hotspots in Turkmenistan, Algeria, Oman, and offshore in the Persian Gulf. We show that oil/gas methane emissions for individual countries are not related to production, invalidating a basic premise in the construction of activity-based bottom-up inventories. Instead, local infrastructure and management practices appear to be key drivers of oil/gas emissions, emphasizing the need for including top-down information from atmospheric observations in the construction of oil/gas emission inventories. We examined the methane intensity, defined as the upstream oil/gas emission per unit of methane gas produced, as a measure of the potential for decreasing emissions from the oil/gas sector, and using as reference the 0.2 % target set by industry. We find that the methane intensity in most countries is considerably higher than this target, reflecting leaky infrastructure combined with deliberate venting or incomplete flaring of gas. However, we also find that Kuwait, Saudi Arabia, and Qatar meet the industry target and thus show that the target is achievable through capture of associated gas, modern infrastructure, and concentration of operations. Decreasing methane intensities across the Middle East and North Africa to 0.2 % would achieve a 90 % decrease in oil/gas upstream emissions and a 26 % decrease of total anthropogenic methane emissions in the region, making a significant contribution toward the Global Methane Pledge. [ABSTRACT FROM AUTHOR]

We quantify weekly methane emissions at 0.25°×0.3125° (≈25×25 km²) resolution from the Permian Basin, the largest oil production basin in the United States, by inverse analysis of satellite observations from the TROPOspheric Monitoring Instrument (TROPOMI) from May 2018 to October 2020. The mean oil and gas emission from the region (± standard deviation of weekly estimates) was 3.7 ± 0.9 Tg a-1, higher than previous TROPOMI inversion estimates that may have used too-low prior emissions or biased background assumptions. We find strong week-to-week variability in emissions superimposed on longer-term trends, and these are consistent with independent inferences of temporal emission variability from tower, aircraft, and multispectral satellite data. New well development and local natural gas spot price were significant drivers of variability in emissions over our study period, but the concurrent 50 % increase in oil and gas production was not. The methane intensity (methane emitted per unit of methane gas produced) averaged 4.6 % ± 1.3 % and steadily decreased over the period from 5–6 % in 2018 to 3–4 % in 2020. While the decreasing trend suggests improvement in operator practices during the study period, methane emissions from the Permian Basin remained high, with methane intensity an order of magnitude above recent industry targets of <0.2 %. Our success in using TROPOMI satellite observations for weekly estimates of emissions from a major oil production basin shows promise for application to near-real-time monitoring in support of climate change mitigation efforts. [ABSTRACT FROM AUTHOR]

The use of satellite instruments to detect and quantify methane emissions from fossil fuel production activities is highly beneficial to support climate change mitigation. Different hyperspectral and multispectral satellite sensors have recently shown potential to detect and quantify point-source emissions from space. The Sentinel-2 (S2) mission, despite its limited spectral design, supports the detection of large emissions with global coverage and high revisit frequency thanks to coarse spectral coverage of methane absorption lines in the shortwave infrared. Validation of S2 methane retrieval algorithms is instrumental in accelerating the development of a systematic and global monitoring system for methane point sources. Here we develop a benchmarking framework for such validation. We first develop a methodology to generate simulated S2 datasets including methane point-source plumes. These benchmark datasets have been created for scenes in three oil and gas basins (Hassi Messaoud, Algeria; Korpeje, Turkmenistan; Permian Basin, USA) under different scene heterogeneity conditions and for simulated methane plumes with different spatial distributions. We use the simulated methane plumes to validate the retrieval for different flux rate levels and define a minimum detection threshold for each case study. The results suggest that for homogeneous surfaces, the detection limit of the proposed S2 methane retrieval ranges from 1000 kgh-1 to 2000 kgh-1, whereas for areas with large surface heterogeneity, the retrieval can only detect plumes in excess of 5000 kgh-1. The different sources of uncertainty in the flux rate estimates have also been examined. Dominant quantification errors are either wind-related or plume mask-related, depending on the surface type. Uncertainty in wind speed, both in the 10-m wind (U10) and in mapping U10 to the effective wind (Ueff) driving plume transport, is the dominant source of error for quantifying individual plumes in homogeneous scenes. For heterogeneous scenes, the surface structure underlying the methane plume affects the plume masking and can become a dominant source of uncertainty. [ABSTRACT FROM AUTHOR]