Monitoring of gas flaring (GF)—the burning of natural gas associated with oil extraction—over most oil- and gas-producing areas is challenging due to high costs or difficult field investigations. As GF contributes to both global warming and air pollution, an up-to-date picture (locations, emissions and trends) of global offshore GF can help countries’ energy decarbonization efforts substantially. Although high-resolution satellite sensors regularly capture high-temperature signals from GF, retrieving spatially explicit information and estimating GF volumes from petabyte images remain challenging. Here we developed a monitoring framework and compiled a 20 m resolution inventory of offshore GF sites by analysing ~8.53 million Sentinel-2 images. A robust model (R2 > 0.99) was established to estimate offshore GF volumes from Sentinel-2 metrics. Our findings reveal that Sentinel-2 can pinpoint global offshore GFs to support scientifically sound decision-making; a vital few (~20%) sites are responsible for >80% of offshore GF volumes, calling for more targeted regulations; and the offshore GF volumes have declined by 26.4% from 2016 to 2021. We conclude that the Zero Routine Flaring Initiative by the Global Gas Flaring Reduction Partnership committing governments and oil companies to end routine flaring by no later than 2030 could be 5 years behind schedule.
Gas flaring (GF) is the controlled burning of excess combustible gases (predominantly methane) that cannot be handled for sale or used for safety, operational or economic reasons1,2. GF exacerbates global warming by substantially contributing greenhouse gases (GHGs) and black carbon3,4,5,6,7, posing environmental risks to community health from the associated noise, light and air pollutants8,9,10,11,12. Also, GF deprives local residents of additional revenues and socio-economic development opportunities by wasting a valuable, environment-friendly energy resource. The World Bank’s Global Gas Flaring Reduction Partnership (GGFR) estimates that ~143.4 billion cubic metres (b.c.m.) of associated petroleum gas (APG), equalling a sales value of US$16.5 billion, has been burned in 2021 (ref. 13). Accordingly, the GGFR has launched an initiative to reach zero routine flaring (ZRF) by 2030. However, monitoring GF in most oil- and gas-producing areas is challenging due to the high expenditure or the inconvenience of field investigations, and consequently, the understandings of the GF distribution, emissions and dynamics are inconsistent, incomplete or even absent14,15. Knowledge gaps are particularly acute for offshore GHGs, which are even more difficult to measure, report and validate than onshore GHGs due to inaccessibility. A comprehensive baseline of global offshore GF at a hyperfine resolution will be valuable to support the effective implementation of the 2030-ZRF Initiative.
Satellites can provide independent worldwide GF monitoring over time and are, therefore, the only objective way of verifying the progress on the GGFR reduction goals. Tremendous breakthroughs have been achieved in drawing a global GF picture using heat and light signals emitted from GFs and recorded by thermal infrared or low-light sensors, such as the Operational Line Scan16, Advanced Along-Track Scanning Radiometer17, Moderate-resolution Imaging Spectroradiometer18, Sea and Land Surface Temperature Radiometer19 and Visible Infrared Imaging Radiometer Suite (VIIRS)20,21,22. All of these efforts support the establishment of National GHG Inventories and the Global Stocktake process under the Paris Agreement23. However, the coarse resolution (~0.3–2.7 km at nadir) is their Achilles’ heel—GFs that are not sufficiently prominent can be missed, and GFs in close proximity could be confused24, leading to uncertainties in subsequent volume estimates and hindering the implementation of targeted regulations25,26. Although not designed for GHG monitoring, the Sentinel-2A/2B Multispectral Instrument (MSI) may have considerable potential for locating exact GFs and monitoring global GF dynamics because of its high-temperature-sensitive shortwave infrared (SWIR) bands with 20 m resolution and a 5 day joint revisit cycle at the Equator27,28. Yet, analysing petabyte-scale images and estimating GF volumes at a fine resolution and down-to-source remains technically challenging.
In this Article, we focused on offshore GF and proposed a detection framework that combines cloud computation and local computation, from which a global inventory of offshore GF sites was established through capitalizing on all offshore Sentinel-2A/2B MSI images during 2015–2021. We further developed a model to estimate the GF volumes from MSI metrics at the global, national and site levels. All these practices not only benchmark the state-of-the-art detailed picture (sources, emissions and trends) of global offshore GF but also allow us to address the following three questions: (1) how reliable is the Sentinel-2 MSI in monitoring global offshore GF; (2) what is the worldwide trend in offshore GF volumes and the distance to the 2030-ZRF goal; and (3) what insight can we gain from Sentinel-2 observations to guide the reduction of global offshore GF? The answers are crucial for bridging the gap between research and policy, and potentially providing a model for decarbonization in other energy-intensive industries.
Using Sentinel-2 to pinpoint global offshore GFs
From ~8.53 million Sentinel-2 MSI images (Supplementary Fig. 1), we enumerated 1,718 offshore GF sites with recurrent GFs (Fig. 1) according to their spectral and positional characteristics (Supplementary Fig. 2, and ‘Detection’ section in Methods). The distribution of the global offshore GF sites is spatially concentrated, with ~72.53% of the sites located in the South China Sea, the Gulf of Guinea, the North Sea, the Persian Gulf, the Gulf of Mexico, the Brazilian coasts and their surroundings. Specifically, offshore GF sites are located in the waters of 59 countries, with the top 10 countries being the United Kingdom (152 sites), the United States (146), China (119), Malaysia (105), Brazil (97), Iran (74), Norway (72), the United Arab Emirates (69), Mexico (66) and Nigeria (63).
According to the high-temperature signals recorded by the Sentinel-2 MSI, we proposed a normalized metric, that is, the GF radiance contribution (GF-RC, ‘Metrics’ section in Methods), to quantify the GF volume of each offshore GF site for a calendar year (‘Estimation’ section in Methods). The country-level GF-RCs from 2016 to 2020 exhibit an excellent linear relationship (R2 = 0.99) with the corresponding reported GF volumes, with a root mean square error of 3.81 × 107 sm3 (0.0381 b.c.m.) (Fig. 2a).
We conducted a double-blind validation to better evaluate the model uncertainties by comparing our estimates for 2021 (Supplementary Table 1) with the GF volumes published after our initial submission. The validation proves the capability of the MSI for quantifying GF volumes. The mean absolute error of the estimates and the reported volumes is 0.0178 b.c.m. at the country level (Extended Data Fig. 1) and 0.0013 b.c.m. at the GF-site level (Extended Data Fig. 2).
Furthermore, an inter-method comparison with the widely used VIIRS Night-Fire and Night-Flare products confirms that MSI can monitor global offshore GF with completeness, spatial explicitness and precision. Sentinel-2 MSI detected at least 330 additional offshore GF sites outside the VIIRS Night-Fire during 2015 and 2020 (Extended Data Fig. 3). This difference in detection is even greater on a monthly scale: ~40% of the sites detected by MSI (Extended Data Fig. 4) are not detected by VIIRS Night-Fire. In addition, the 20 m MSI imagery can better differentiate GFs that are close to each other: 1,370 MSI GF sites are resolved from 1,014 coincident VIIRS sites. Furthermore, the MSI model dominates the VIIRS Night-Flare model, with a higher coefficient of determination (0.99 versus 0.97) and a lower root mean square error value (0.0381 b.c.m. versus 0.0667 b.c.m.) (Fig. 2), and the mean absolute error from the double-blind validations is approximately 30% of the VIIRS results (0.0178 b.c.m. versus 0.0588 b.c.m.) at the country level and 40% at the GF-site level (0.0013 b.c.m. versus 0.0033 b.c.m.).
We, therefore, ascertain that the Sentinel-2 MSI, through an elaborated metric design and calibration and an innovated data processing, can lay the groundwork for the national and global reporting to underpin offshore GF mitigation policy. The capabilities of Sentinel-2, which policymakers may not constantly be aware of, have access to, or fully utilize, will be critical to improving awareness of GF emissions and enabling science-based decision-making29. We also suggest that the high spatial resolution of MSI and the high temporal resolution of VIIRS can be complementary for monitoring the dynamics of remote sites and facilitating informed regulation. For details on the datasets, GF characteristics, reported volumes, detection parameters, metric design, double-blind validation, inter-method comparisons and limitations, please refer to Supplementary Notes1–9.
Steady decline of global offshore GF volume
On the global scale, offshore GF activities exhibited a marked declining trend from 2016 to 2021 in all metrics derived from time-series cloud-free MSI images (‘Metrics’ section in Methods), including the occurrence rates (ORs), mean radiances (MRs) and GF-RCs. Among all offshore GF sites that were active each year from 2016 to 2021, (1) the mean OR exhibited a sustained decline (~6.26%), from 62.34% in 2016 to 56.09% in 2021 (Fig. 3a); (2) the mean MR declined by 24.48% (~180.61 W m−2), from 737.90 W m−2 sr−1 µm−1 in 2016 to 557.29 W m−2 sr−1 µm−1 in 2021 (Fig. 3a); and (3) the total GF-RC declined by ~26.37%, from 883,164 W m−2 sr−1 µm−1 in 2016 to 650,264 W m−2 sr−1 µm−1 in 2021.
The total offshore APG burned from 2016 to 2021 was estimated to be ~150.10 b.c.m., equivalent to ~US$17.23 billion assuming a gas sales value of US$2.5 per Metric Million British Thermal Unit, and a gas heating value of 1,300 British Thermal Unit per standard cubic foot13. The global offshore GF declined steadily from 28.65 b.c.m. in 2016 to 21.16 b.c.m. in 2021, with an annual reduction rate of 1.49 b.c.m. (Fig. 3b). Assuming that GF sites with ORs ≤0.1 are intermittent GF (~0.31% of the total in 2021, ~0.07 b.c.m.) and the reduction rate (1.49 b.c.m. per year) continues, the offshore GF volume will decline to 0.44 b.c.m. by the end of 2035, approaching the ZRF goal (Fig. 3b), at least 5 years behind the scheduled 2030 GGFR goal. An independent prediction based on VIIRS offshore GF estimates from 2016–2021 exhibits the same results (Fig. 3b).
We further estimated the offshore GF volumes by country from 2016 to 2021 (Supplementary Table 1). A Mann–Kendall test shows that most countries have substantially curtailed offshore GF. Among the 59 countries that burn offshore APG, ~51% demonstrate a notable or some degree of decrement during 2016–2021 (Fig. 4), and the countries that burned the most tended to reduce the most—8 of the top 10 APG-burning countries contributed ~76.27% of the total reduction (9.06 b.c.m.). In particular, the reductions in Nigeria (ISO code: NGA, 1.78 b.c.m.), Angola (AGO, 1.71 b.c.m.), the Democratic Republic of the Congo (COD, 1.08 b.c.m.) and Malaysia (MYS, 1.03 b.c.m.) exceeded 1 b.c.m. In contrast, the number of countries with a noticeable or certain upward trend is much smaller than the number of countries with a downward trend (8 versus 30). The top seven countries with the utmost increments, namely Ghana (GHA, 0.45 b.c.m.), China (CHN, 0.25 b.c.m.), Guyana (GUY, 0.23 b.c.m.), India (IND, 0.15 b.c.m.), the United Arab Emirates (ARE, 0.14 b.c.m.), Australia (AUS, 0.08 b.c.m.) and Saudi Arabia (SAU, 0.07 b.c.m.), contributed 87.89% of the total increment (1.57 b.c.m.) from 2016 to 2021.
Implications of the 80/20 Rule
The performance of individual offshore GF sites differs markedly—the GF-RCs of 1,718 GF sites from 2016–2021 range over several orders of magnitude (0.61–144,465.21 W m−2 sr−1 µm−1), and the same vast disparities of ORs and MRs (Supplementary Figs. 3 and 4). Statistically, their distribution follows the 80/20 Rule: for many outcomes, roughly 80% of the consequences come from 20% of the causes. From the OR and MR perspectives (Fig. 5), a minority of sites produce most of the total GF-RCs: GF sites together with the uppermost 20% of ORs and the uppermost 20% of MRs (175 sites, red box in Fig. 5b), account for ~58.4% of the total GF-RCs; and 26.5% of sites (yellow box in Fig. 5b) for 82.0% of the total. From the GF-RC perspective, the 80/20 Rule is more prominent: ~80% of the total GF-RCs come from the vital few (the 356 sites with the uppermost GF-RCs, accounting for ~20.72% of all GF sites), whereas the rest come from the trivial many (the remnant 1,362 sites). For an extreme example, the site with the largest GF-RC (in the Iranian waters) is greater than the sum (143,867.12 W m−2 sr−1 µm−1) of the 880 sites with the smallest GF-RCs.
For policymakers, the revealed 80/20 Rule of individual GF sites can better support the introduction of targeted regulations on the vital few offshore GF sites for global and national GF mitigation. Furthermore, at the operational level, the down-to-source MSI inventory can offer targeted information on the location and severity of GF sites and how much additional revenue can be generated by mitigating offshore GF, which may not be fully recognized by oil and gas operators. Admittedly, APG monetization requires a considerable investment and substantial carrying costs30,31; according to the International Energy Agency, most gas-utilization technologies will be uneconomic when APG volumes are <10,000 m3 per day32. Nonetheless, our estimate (Extended Data Fig. 5) shows that 820 offshore GF sites, burning ~96.53% of the global offshore GF volumes from 2016 to 2021, can balance these costs and revenues (burning APG >0.02191 b.c.m. from 2016 to 2021), and 172 sites burning ten times the limit (>0.2191 b.c.m. from 2016 to 2021), accounting for 61.52% of the global offshore GF volumes, are the most promising. Most of the 172 sites are located in the Gulf of Guinea (65 sites), the South China Sea (31 sites), the Persian Gulf (20 sites) and their surrounding waters. Focusing limited funds and efforts on the vital few offshore GF sites should therefore be practical and effective, especially in the context of record natural gas prices caused by ongoing geopolitical tensions between Russia and Ukraine.
The estimated total offshore GF volume of 2016–2021 (~150.10 b.c.m.) equals 420.28 million tonnes of CO2 equivalent (CO2e) emissions; if each cubic metre of a typical-composition APG flared results in ~2.8 kg of CO2e emissions at a flare combustion efficiency of 98% and a global warming potential for methane of 25 (ref. 13). Correspondingly, the average annual offshore GF volume is ~70 million tonnes of CO2e, representing 2.0‰ of global CO2 emissions from fossil-fuel combustion in 2020 (34.8 billion tonnes)33. Although the number appears small, it is important because (1) the global anthropogenic carbon emissions consist of many trivial subcategories of emissions, and each category matters; (2) GF is completely unproductive and can be avoided far more easily than emissions from other industrial sectors13; (3) if properly managed, the APG that is burned could otherwise be used to partially replace fuels (for example, coal) that have higher emissions per unit of energy, further reducing CO2 emissions.
Our study shows a substantial reduction in offshore GF from 2016 to 2021. Even so, at the current rate of decline, our study suggests that the GGFR 2030-ZRF goal will be delayed until at least 2035. The decline could be caused by active mitigation measures or passive adaption to market demand, although the decisive factor in this decline remains unknown. Notably, brisk demands for oil will beget more GF and vice versa—these demands are closely related to global economic growth, which remains substantially uncertain in the post-COVID-19 era. Nevertheless, given that offshore GF volumes are usually unpublished, incomplete or even unavailable for various reasons, Earth observation technologies, such as the previous VIIRS model and the MSI model developed here, could enhance data transparency and help to adjust the countermeasures to curb global offshore GF in a timely and appropriate manner.
The proposed GF monitoring framework relies on the radiance emitted by high-temperature sources and, therefore, cannot be applied to estimate the direct release of APG into the atmosphere, namely venting. Previous studies have demonstrated the presence of super/ultra methane emitters in oil and gas development areas34,35,36. In future work, the combined utilization of space-borne GHG products derived from the Tropospheric Monitoring Instrument, Orbiting Carbon Observatory-2/3 or GHGSat will enhance the understanding of flaring and venting emissions in the offshore oil/gas development industry.
Lastly, decarbonizing the world’s energy systems is an essential pillar of global action on carbon neutrality. This study focused on offshore GF because offshore GF is a crucial component of global GF, and the delay of the subgoal will undoubtedly affect the achievement of the overall GHG target. Furthermore, the success of global offshore GF applications could shed light on the monitoring of onshore industrial heat sources. Admittedly, pinpointing onshore GF is needed to verify the complete progress of reduction goals, which is ongoing and needs time because (1) numerous onshore false positives (for example, Supplementary Figs. 5 and 6) need to be filtered out with the aid of high-resolution images and carefully designed algorithms; (2) the number of global onshore GF sites we detected is substantially greater than the currently reported; and (3) heterogeneous onshore backgrounds require further calibration between MSI metrics and reported volumes. Even so, our further experiments show that the MSI framework is promising for monitoring high-temperature sources in energy-intensive industries, such as GFs in shale oil/gas pads37, kilns in cement plants and blast furnaces in steel plants28. The prospect offered by Sentinel-2-like satellites will support actions in the race to net zero emissions going forward on the local, national and global levels.
We proposed a monitoring framework (Extended Data Fig. 6) that combines the Google Earth Engine cloud computation38 and local computation to process global offshore MSI images and to understand the distribution, dynamics and GF volumes of global offshore GF sites. Specifically, the framework includes three modules, that is, offshore GF site detection, MSI metrics analysis and offshore GF volume estimation. This framework is particularly suitable for analysing the dynamics of extremely small and sparsely distributed geo-features worldwide. For more details on the framework, please refer to Supplementary Note 6.
The offshore GF has an unusually high temperature, generally ranging from 1,275 K to 2,275 K (ref. 39), and results in an emission peak and substantial increase of the spectral radiance in the MSI SWIR bands (the MSI near-SWIR, Band 11, ~1.60 μm, and the far-SWIR, ~2.20 μm)28. Moreover, repeated GFs from a specific burner can be detected frequently in time-series MSI images with excellent geometric accuracy40 and therefore are densely distributed over relatively small areas (Supplementary Note 2). According to the spectral and positional characteristics of offshore GFs, we used the following three steps to detect offshore GF sites where GFs occur repeatedly. (1) We adopted the thermal anomaly index method28 to detect high-temperature anomalies from single-phase MSI Top-Of-Atmosphere reflectance images according to their spectral characteristics; (2) we refined offshore GF site candidates according to the higher occurrence frequency of repeated GFs among the time-series detections to mitigate false positives, such as those caused by the sun glinting over water, moving clouds along the seam lines of two neighbouring granules, or cloud edges28; and (3) we utilized auxiliary data, including land mask data and the world volcano database, to exclude onshore false positives and offshore active volcanos. We then selected offshore GF sites adjacent to the global offshore oil/gas platform inventory we built previously40,41,42. For those unconformities, careful visual inspections were conducted. We further assigned the country names and the ISO codes to the individual sites according to their location in the corresponding exclusive economic zones. Extended Data Fig. 6 gives all the parameters used to detect offshore GF sites from MSI imagery. A detailed explanation of the detection of offshore GF sites is given in Supplementary Note 4.
Over a given period (for example, one calendar year), the total volume of APG combusted at each GF site is the product of the burning time and the mean burning volume per time unit. In contrast, the instantaneous APG volume burned at an individual GF site varies over time, depending on the air–fuel ratio, the mass flux of the flue gas, burner diameter and ambient atmospheric conditions (for example, atmospheric temperature, pressure and wind speed/direction)2.
From the perspective of dense time-series MSI observations at a given GF site over one calendar year, the detected number of GFs can approximate the burning time, and the MR emitted by GFs and recorded in the high-temperature-sensitive MSI bands can approximate the mean burning volumes. Admittedly, the spatiotemporal availability of MSI images is highly variable (Supplementary Fig. 1) and dramatically affected by clouds, making it challenging to compare GF estimates among offshore GF sites and years directly. Therefore, we propose the following three normalized metrics, calculated from a time series of cloud-free MSI observations, to approximate the associated activities (that is, burning time, mean burning volume and total burned volume) of each offshore GF site.
(1) The OR describes the degree of frequency of GFs observed from the time series of MSI observations, that is, the ratio of the number of GFs detected to the number of MSI observations. (2) The MR describes the average intensity of the GF using the radiance of the MSI SWIR and NIR bands (bands 12, 11 and 8A) averaged over one year for all detected cases of a GF site. (3) The GF-RC is the product of the OR and the MR to quantify the flare volume of each offshore GF site. For each offshore GF site, the above metrics were calculated from time-series cloud-free MSI images over each calendar year. A graphical representation (Supplementary Fig. 7) was given to show the detailed calculation. The GF-RCs at the flaring site level were then aggregated by country and year to upscale to the country level. For more details on the design of the MSI metrics, please refer to Supplementary Note 5.
Despite the widespread occurrence of offshore GF in 59 countries, only six countries (Brazil, Denmark, the Netherlands, Norway, the United Kingdom and the United States) provide publicly accessible offshore GF volumes. We paired these data with the annual GF-RCs of the corresponding GF sites and obtained 429 data pairs from 2016 to 2020 (Supplementary Note 2). To avoid biasing the estimates towards countries with more data, we aggregated the 429 data pairs by country and year, resulting in 25 data pairs at a larger aggregated level (the country level). The derived country-level MSI model is given in Fig. 2a.
We use the MSI model to estimate the annual offshore GF volumes by country, according to the corresponding GF-RCs aggregated at the country level. The annual global offshore GF volumes are the sum of the country estimates for the corresponding year. Moreover, at the platform/field level, the data pairs between GF-RCs and reported flared volumes also show a strong linear relationship (Extended Data Figs. 7 and 8). We, therefore, allocate the annual national GF volumes to individual GF sites according to their GF-RC share in the respective country.
This study focused on pinpointing global offshore GF from Sentinel-2 MSI imagery, including detecting offshore GF sites, the paring between MSI metrics and government-reported data, and estimating flaring volumes. Given the highly variable nature of GFs, the MSI imagery and the various factors influencing the detection/estimation accuracy, the derived global picture is not immune from limitations and uncertainties. We performed an uncertainty analysis on multiple factors, including the spatial coverage and temporal resolution of the MSI images, the quality of the quality assessment band of the MSI images, the saturation of the MSI bands, the missing value pixels in GF regions, the reported GF volumes, the GF determination and the background contamination. The results of this uncertainty analysis are presented in Supplementary Note 9. Even with the above uncertainties, we argue that the double-blind validation and the inter-method comparison have demonstrated the robustness of the MSI model, as analysed in Supplementary Notes 7 and 8.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
The codes for detecting high-temperature anomalies and calculating related MSI metrics are available on GitHub (https://github.com/yongxuenju/yongxue_code).
Emam, E. A. Gas flaring in industry: an overview. Pet. Coal 57, 532–555 (2015).
Fawole, O. G., Cai, X.-M. & MacKenzie, A. Gas flaring and resultant air pollution: a review focusing on black carbon. Environ. Pollut. 216, 182–197 (2016).
Amaechi, C. F. & Biose, E. Gas flaring: carbon dioxide contribution to global warming. J. Appl. Sci. Environ. Manag. 20, 309–317 (2016).
Caseiro, A., Gehrke, B., Rücker, G., Leimbach, D. & Kaiser, J. W. Gas flaring activity and black carbon emissions in 2017 derived from the Sentinel-3A Sea and Land Surface Temperature Radiometer. Earth Syst. Sci. Data 12, 2137–2155 (2020).
Conrad, B. M. & Johnson, M. R. Field measurements of black carbon yields from gas flaring. Environ. Sci. Technol. 51, 1893–1900 (2017).
Ismail, O. S. & Umukoro, G. E. Global impact of gas flaring. Energy Power Eng. 4, 290–302 (2012).
Johnson, M., Devillers, R. & Thomson, K. A generalized sky-LOSA method to quantify soot/black carbon emission rates in atmospheric plumes of gas flares. Aerosol Sci. Technol. 47, 1017–1029 (2013).
Allshouse, W. B., McKenzie, L. M., Barton, K., Brindley, S. & Adgate, J. L. Community noise and air pollution exposure during the development of a multi-well oil and gas pad. Environ. Sci. Technol. 53, 7126–7135 (2019).
Anejionu, O. C., Whyatt, J. D., Blackburn, G. A. & Price, C. S. Contributions of gas flaring to a global air pollution hotspot: spatial and temporal variations, impacts and alleviation. Atmos. Environ. 118, 184–193 (2015).
Cushing, L. J., Vavra-Musser, K., Chau, K., Franklin, M. & Johnston, J. E. Flaring from unconventional oil and gas development and birth outcomes in the Eagle Ford Shale in South Texas. Environ. Health Perspect. 128, 077003 (2020).
Jacobson, M. Z. Short-term effects of controlling fossil-fuel soot, biofuel soot and gases, and methane on climate, Arctic ice, and air pollution health. J. Geophys. Res. 115, D14209 (2010).
Motte, J., Alvarenga, R. A., Thybaut, J. W. & Dewulf, J. Quantification of the global and regional impacts of gas flaring on human health via spatial differentiation. Environ. Pollut. 291, 118213 (2021).
Global Gas Flaring Tracker Report. World Bank Global Gas Flaring Reduction Partnership https://www.worldbank.org/en/programs/gasflaringreduction/global-flaring-data (2021).
Faruolo, M., Caseiro, A., Lacava, T. & Kaiser, J. W. Gas flaring: a review focused on its analysis from space. IEEE Geosci. Remote Sens. Mag. 9, 258–281 (2020).
Zhang, Z., Sherwin, E. D. & Brandt, A. R. Estimating global oilfield-specific flaring with uncertainty using a detailed geographic database of oil and gas fields. Environ. Res. Lett. 16, 124039 (2021).
Elvidge, C. D. et al. A fifteen year record of global natural gas flaring derived from satellite data. Energies 2, 595–622 (2009).
Casadio, S., Arino, O. & Serpe, D. Gas flaring monitoring from space using the ATSR instrument series. Remote Sens. Environ. 116, 239–249 (2012).
Anejionu, O. C., Blackburn, G. A. & Whyatt, J. D. Detecting gas flares and estimating flaring volumes at individual flow stations using MODIS data. Remote Sens. Environ. 158, 81–94 (2015).
Fisher, D. & Wooster, M. J. Multi-decade global gas flaring change inventoried using the ATSR-1, ATSR-2, AATSR and SLSTR data records. Remote Sens. Environ. 232, 111298 (2019).
Elvidge, C. D., Zhizhin, M., Hsu, F.-C. & Baugh, K. E. VIIRS Nightfire: satellite pyrometry at night. Remote Sens. 5, 4423–4449 (2013).
Elvidge, C. D., Zhizhin, M., Baugh, K., Hsu, F.-C. & Ghosh, T. Methods for global survey of natural gas flaring from Visible Infrared Imaging Radiometer Suite data. Energies 9, 14 (2015).
Liousse, C. et al. Flaring emissions in Africa: distribution, evolution and comparison with current inventories. Atmos. Environ. 199, 423–434 (2019).
Tollefson, J. ‘Flaring’ wastes 3.5% of world’s natural gas. Nature https://doi.org/10.1038/nature.2016.19141 (2016).
Anejionu, O. C., Blackburn, G. A. & Whyatt, J. D. Satellite survey of gas flares: development and application of a Landsat-based technique in the Niger Delta. Int. J. Remote Sens. 35, 1900–1925 (2014).
Anejionu, O. C. Rationale, historical developments and advances in remote sensing of gas flares. Int. J. Remote Sens. 40, 6700–6719 (2019).
Brandt, A. R. Accuracy of satellite-derived estimates of flaring volume for offshore oil and gas operations in nine countries. Environ. Res. Commun. 2, 051006 (2020).
Berger, M., Moreno, J., Johannessen, J. A., Levelt, P. F. & Hanssen, R. F. ESA’s Sentinel missions in support of Earth system science. Remote Sens. Environ. 120, 84–90 (2012).
Liu, Y., Zhi, W., Xu, B., Xu, W. & Wu, W. Detecting high-temperature anomalies from Sentinel-2 MSI images. ISPRS J. Photogramm. Remote Sens. 177, 174–193 (2021).
GHG monitoring from space: a mapping of capabilities across public, private, and hybrid satellite missions. GEO, ClimateTRACE, WGIC https://earthobservations.org/documents/articles_ext/GHG%20Monitoring%20from%20Space_report%20final_Nov2021.pdf (2021)
Buzcu-Guven, B. & Harriss, R. Extent, impacts and remedies of global gas flaring and venting. Carbon Manag. 3, 95–108 (2012).
Mansoor, R. & Tahir, M. Recent developments in natural gas flaring reduction and reformation to energy-efficient fuels: a review. Energy Fuels 35, 3675–3714 (2021).
Optimising Russian natural gas: reform and climate policy. The Organisation for Economic Co-operation and Development, Paris, France. International Energy Agency https://www.iea.org/reports/optimising-russian-natural-gas (2006)
The Global Carbon Project’s fossil CO2 emissions dataset. Global Carbon Project https://www.globalcarbonproject.org/carbonbudget/21/data.htm (2021).
Duren, R. M. et al. California’s methane super-emitters. Nature 575, 180–184 (2019).
Lauvaux, T. et al. Global assessment of oil and gas methane ultra-emitters. Science 375, 557–561 (2022).
Plant, G. et al. Inefficient and unlit natural gas flares both emit large quantities of methane. Science 377, 1566–1571 (2022).
Wu, W. et al. Monitoring gas flaring in Texas using time-series Sentinel-2 MSI and Landsat-8 OLI images. Int. J. Appl. Earth Obs. Geoinf. 114, 103075 (2022).
Gorelick, N. et al. Google Earth Engine: planetary-scale geospatial analysis for everyone. Remote Sens. Environ. 202, 18–27 (2017).
Liu, Y. et al. Identifying industrial heat sources using time-series of the VIIRS Nightfire product with an object-oriented approach. Remote Sens. Environ. 204, 347–365 (2018).
Liu, Y. et al. Geometric accuracy of remote sensing images over oceans: the use of global offshore platforms. Remote Sens. Environ. 222, 244–266 (2019).
Xu, W. et al. Proliferation of offshore wind farms in the North Sea and surrounding waters revealed by satellite image time series. Renew. Sust. Energ. Rev. 133, 110167 (2020).
Liu, Y. et al. Assessment of offshore oil/gas platform status in the northern Gulf of Mexico using multi-source satellite time-series images. Remote Sens. Environ. 208, 63–81 (2018).
We thank Google Earth Engine for providing Sentinel-2 MSI data and online processing. We are especially grateful to K. E. Rosendahl of the Norwegian University of Life Science for providing the Norwegian flaring volume data. We thank the Earth Observation Group, Payne Institute for Public Policy, Colorado School of Mines for the VIIRS Night-Fire (VNF) nightly data. We also sincerely thank all data providers listed in Supplementary Table 3 for their continuous efforts and data sharing. Y.L. acknowledges funding from the Key Research and Development Program of China (grant no. 2019YFA0606601). Y.Z. acknowledges funding from the National Natural Science Foundation of China (42007198).
The authors have no competing interests to declare.
Peer review information
Nature Sustainability thanks Johannes Kaiser and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended Data Fig. 1 Differences between the MSI/VIIRS estimates and the reported offshore GF volumes of Denmark, the United Kingdom, the northern GOM of the United States of America, and the Netherlands (data unavailable for 2021) of 2016–2021 at the country level.
The red points represent the double-blind comparison for 2021. We excluded Brazil and Norway here because Brazil only reports volumes of the largest twenty offshore GF oil/gas fields, and we only obtained one-year data from Norway. The bars above the horizontal axis represent estimates greater than the reported volumes and vice versa.
Extended Data Fig. 2 Double-blind validations on the MSI (a–c) and VIIRS GF volume estimates (a1–c1) for 2021 at the GF site level.
a, a1, for Denmark; b, b1, for the United Kingdom; c, c1, for the northern GOM of the United States of America. The 2021 data pairs (purple symbols) are for double-blind validation and were excluded from the model establishment. The green symbols represent data pairs with reported volumes but without corresponding VIIRS estimates and therefore are excluded from fitting and calculating the mean absolute error (MAE). The dark and light red shadows represent the 95% fitting and predicted confidence, respectively.
Extended Data Fig. 3 Comparison between offshore GF sites derived from time-series MSI images and VIIRS Night-fire products.
1a–5e, Zoom-in insets of the northern GOM, southern GOM, the North Sea, Persian Gulf, Gulf of Guinea, and the South China Sea and their surrounding waters. The green points represent GF sites that both MSI and VIIRS identified, and the red points were GF sites identified by VIIRS but omitted by MSI, whereas the yellow points were GF sites identified by MSI but omitted by VIIRS. Source: VIIRS Nightfire, Colorado School of Mines.
Extended Data Fig. 4 Monthly comparison between offshore GF sites derived from time-series MSI images and monthly VIIRS Night-Fire (VNF) products.
a, The number (purple dots) of GF sites detected by the VIIRS VNF product per month. Dark blue: the number of offshore GF sites detected from VIIRS VNF products that coincided with those from MSI images; light blue: the number of GF sites omitted by MSI images but detected by the VIIRS VNF product. b, The number (green dots) of GF sites detected by MSI images per month. Dark green: the number of offshore GF sites detected from MSI images and they coincided with those from the VIIRS VNF product; light green: the number of GF sites detected by MSI images but omitted by the VIIRS VNF product.
Offshore GF sites with a high potential (red points), moderate potential (yellow points), and low potential (green points) for monetisation.
Global offshore GF monitoring framework based on time-series Sentinel-2 MSI images.
Extended Data Fig. 7 Relationships between the MSI GF-RCs/VIIRS estimates and the reported offshore GF volumes at the platform-/field level.
a, Relationships between GF volumes and GF-RCs at the platform-/field level. b, Post-assessment on VIIRS estimates, using the same GF volumes. The GF-RCs were calculated from time-series cloud-free MSI observations. Six countries independently reported the GF volumes. The 2021 data pairs are for double-blind validation and therefore were excluded from the fitting analysis. The dark and light red shadows represent the 95% fitting and predicted confidence, respectively.
Extended Data Fig. 8 Coincidence between VIIRS GF volume estimates and MSI estimates from 2016 to 2020.
a, the MSI GF-RC and the VIIRS estimates at the platform-/field level. b, the MSI and VIIRS estimate at the platform-/field level. c, the MSI GF-RCs and the VIIRS estimate at the country level. d, the MSI and VIIRS estimate at the country level.
About this article
Cite this article
Liu, Y., Pu, Y., Hu, X. et al. Global declines of offshore gas flaring inadequate to meet the 2030 goal. Nat Sustain 6, 1095–1102 (2023). https://doi.org/10.1038/s41893-023-01125-5