-
Created by
Michela Giusti, last updated on Jul 02, 2026
10 minute read
Contributors: I Aben, S Kurchaba, AI Lopez-Norena, THP Newman, KL Louzada, and JD Maasakkers
History of modifications
Acronyms
Introduction
CAMS Methane Hot Spot service shows an overview of methane plumes from methane super-emitters detected using data from the TROPOspheric Monitoring Instrument (TROPOMI) onboard the Copernicus Sentinel-5 Precursor satellite. The presented plumes are detected using an updated version of the two-step machine learning pipeline, following the methodology described in Schuit et al. (2023), and manually verified by two experts. The map available through the CAMS Methane Hotspot Explorer shows approximate source locations based on single TROPOMI plumes and initial source rate estimates calculated using the Integrated Mass Enhancement (IME) method. Due to factors such as varying emissions, cloud cover, and viewing geometry, the number of detected plumes fluctuates from one week to another. The provided detections focus on plume-like structures and thus exclude larger-scale enhancements such as those seen over the Permian Basin or over wetland areas. While other source sectors, such as livestock, are responsible for a large fraction of total emissions, those emissions are not spatially concentrated enough to be detected as individual plumes with TROPOMI and therefore do not show up on the Methane Explorer.
Please note that delivered plumes have only been subject to initial verification, precise quantification and final interpretation require more extensive evaluation.
1. TROPOMI observations
The TROPOspheric Monitoring Instrument (TROPOMI) (Veefkind et al., 2012) on board ESA's Sentinel-5P satellite was launched in 2017. Among others, it observes atmospheric dry-air methane column mixing ratios with a pixel size down to 7 km × 5.5 km and daily global coverage, resulting in a point source detection limit down to ∼8 t h−1 (Schuit et al., 2023). Detections are made using operational TROPOMI Level-2 methane data product available from the Copernicus Data Space.
2. Source location estimation
For each detected plume, we estimate its source location based on the most upwind TROPOMI pixel included in the detected plume (Schuit et al., 2023). Because of TROPOMI's kilometer-scale resolution and uncertainty in the inclusion of pixels in the plume, this estimated location does usually not exactly align with the facility responsible for the detected plume. For transient emissions, where plumes can disconnect from their sources and drift through the atmosphere, the distance between the reported source location estimate and the true source can be larger.
3. Source rate quantification
The plumes' source rates are quantified using the Integrated Mass Enhancement (IME) method which integrates the observed methane enhancement in the plume with an effective wind speed derived from an ensemble of reanalysis wind data (Frankenberg et al., 2016; Varon et al., 2018) using ECMWF ERA5 10-m winds (Hersbach et al., 2020), GEOS FP 10-m winds, and GEOS FP planetary boundary layer (PBL) winds (Molod et al., 2012). The emission rate (Q) is calculated as follows:
where τ is the plume's atmospheric lifetime, Ueff is the effective wind speed, L is the plume length, and ∆Ω denotes the methane column mass enhancement above the local background of pixel j, with the footprint Aj for N pixels in the plume (Schuit et al., 2023).
Reanalysis wind data from GEOS-FP is obtained from portal.nccs.nasa.gov and 10-m wind data from ECMWF as included in the TROPOMI level-2 methane data product.
4. Source type classification
Because of the limited spatial resolution of TROPOMI (7 km x 5.5 km), the exact sources underlying detected plumes can usually not be identified. We therefore report a most likely source type based on the relative size of different bottom-up emission products at the location of the detected plume. For a given plume, the reported source type is determined based on gridded bottom-up emission inventories in a 0.7 × 0.7 [deg] square centered around the estimated source location, reporting the largest source sector. Schuit et al. (2023) found that using a window of this size mitigates errors in the estimated source location and spatial errors in the emission inventories.
As bottom-up emission inventories, we use 2019 oil, gas, and coal emissions from the updated Global Fuel Exploitation Inventory (GFEI v2; Scarpelli et al., 2022) and 2018 solid waste emissions from EDGAR V6.0 (Crippa et al., 2021). We do not include sectors that have a low probability to produce point source emissions that can be detected in single overpass TROPOMI data, such as rice cultivation and livestock.
Following Schuit et al. (2023), we also do not use this approach to attribute detections to wetlands. Nevertheless, we study the fluxes from the 2019 WetCHARTs v1.3.1 ensemble (Bloom et al., 2021) to identify regions where detections might be influenced by strong wetland fluxes, as, for example, in central Africa (Pandey et al., 2021). Such detections over Africa (Based on ArcGIS continent maps) are reported under the category “Unclassified”, as apart from wetlands there could potentially be other emission sources that are not accounted for in the emission inventories but are contributing to the detected plumes.
5. Update to the Schuit et al. (2023) methodology
The operational machine learning pipeline was updated to v1.2 for detections made on and after June 21, 2026. This version introduces updates to two key components of the methodology described in Schuit et al. (2023).
Destriping Algorithm Optimisation. The first update concerns the destriping algorithm applied to TROPOMI observations. Destriping is performed using the moving-median smoothing approach described by Borsdorff et al. (2024). In v1.2, the smoothing parameters were optimised to reduce stripe artefacts while improving the detection rate of true methane plumes.
In pipeline v1.1, the destriping algorithm used smoothing windows of 7 pixels across the flight path and all available pixels along flight path for each orbit. In v1.2, these parameters were updated to 14 pixels across-path and 100 pixels along-path. The optimisation was performed over a dataset of 1,200 TROPOMI orbits from 2025, consisting of the first 100 orbits from each month.
Second-Stage Classifier Upgrade. The second update concerns the second-stage classifier. The Support Vector Classifier used in v1.1 has been replaced by an XGBoost classifier in v1.2. The XGBoost model (Chen et al., 2016) was trained using additional data collected and labelled for the CAMS Methane Hotspot Service.
Performance Evaluation. Operational performance was evaluated by comparing pipeline versions v1.1 and v1.2 over a nine-week period in 2026 (weeks 13–21). The evaluation showed that v1.2 achieved an increase in mean precision from 0.48 to 0.65 (+35%) and an increase in the mean number of detected plumes from 27.6 to 38.1 (+38%). These improvements indicate that v1.2 detects a greater number of true methane plumes while reducing the proportion of false positive detections requiring manual verification by experts.
Detections before June 22, 2026 will not be preprocessed using the new algorithm. As the updates can lead to the improved detection of methane plumes, users should be careful interpreting trends in the number of detected plumes over time.
Data organization, access and format
An interactive overview of the detected methane plumes is available through the methane-explorer application.
Figure 1: Methane Hotspots Explorer map.
The complete file of methane plumes detected since the starting date of the project (1 May 2024) is available as a single CSV file. The file is updated on a weekly basis.
The methane-explorer also provides detailed detection information along with a visual overlay of the column-averaged methane mixing ratios measured by TROPOMI and the plume mask used as input for estimating the source emission rate using the Integrated Mass Enhancement (IME) method. It also includes near-surface wind data used to estimate the plume’s source location and emission rate.
Spatial grid
The delivered methane plumes are detected based on TROPOMI/S5P data with the pixel size down to 7 km x 5.5 km at nadir. Source location estimates are reported with a resolution of 0.01, but have significant uncertainty as described under 'Source location estimation'.
Temporal frequency
The TROPOMI instrument used for the detection of methane plumes has daily global coverage. Only clear-sky observations can be used to detect methane plumes.
Data update frequency
Detected plumes are uploaded on a weekly basis (starting from 1 May 2024). The number of detections fluctuates from week to week because of varying emissions, cloud cover, and viewing geometry.
Output data file content
Table 1, provides the available output data for each methane plume detection (also in the CSV file) :
Table 1: Output data per detected plume
| Name | Description | Units |
|---|---|---|
| date | Observation date |
YYYYMMDD |
| time_UTC | Observation time | HH:MM:SS |
| lat | Estimates source location latitude | degrees |
| lon | Estimates source location longitude | degrees |
| source_rate_t/h | Estimated emission rate based on the emission quantification ensemble | tonnes per hour |
| uncertainty_t/h | Uncertainty of the emission rate based on the emission quantification ensemble | tonnes per hour |
| source_type | Likely emission source type | Coal, Oil, Gas, Landfill (urban), and Unclassified |
| source_country | *Country where the emission was detected |
*The attribution to a specific country was performed based on ArcGIS country maps, accessed: February 2024.
Guidelines
Unavailability or delay of the NOAA's Suomi-NPP VIIRS cloud data used in the TROPOMI level-2 methane data product may result in the delay of the delivery of methane plumes. In the production of the TROPOMI level-2 methane data product, there is a 3-day waiting period for the VIIRS cloud data to become available. After that, if the VIIRS cloud data is still absent, an alternative cloud product based solely on TROPOMI observations is used (Borsdorff et al., 2024). As a result, the detection of plumes based on these orbits can be delayed with results reported in the next week.
Data before June 21, 2026 were processed with v1.1 of the plume detection pipeline and data on and after using v1.2 (see Introduction, Section 5). As the updates can lead to the improved detection of methane plumes, users should be careful interpreting trends in the number of detected plumes over time.
References
Bloom, A., Bowman, K., Lee, M., Turner, A., Schroeder, R., Worden, J., Weidner, R., McDonald, K., and Jacob, D.: CMS: Global 0.5-deg Wetland Methane Emissions and Uncertainty (WetCHARTs v1.3.1), ORNL DAAC [data set], https://doi.org/10.3334/ORNLDAAC/1915, 2021.
Borsdorff, T.; Martinez-Velarte, M.C.; Sneep, M.; ter Linden, M.; Landgraf, J. Random Forest Classifier for Cloud Clearing of the Operational TROPOMI XCH4 Product. Remote Sens., 16, 1208. https://doi.org/10.3390/rs16071208, 2024.
Chen, T., and Guestrin, C.: XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, https://doi.org/10.1145/2939672.2939785, 2016.
Crippa, M., Guizzardi, D., Muntean, M., Schaaf, E., Lo Vullo, E., Solazzo, E., Monforti-Ferrario, F., Olivier, J., and Vignati, E.: EDGAR v6.0 Greenhouse Gas Emissions [Dataset], European Commission, Joint Research Centre (JRC) [data set], http: //data.europa.eu/89h/97a67d67-c62e-4826-b873-9d972c4f670b (last access: 20 April 2023), 2021.
Frankenberg, C., Thorpe, A. K., Thompson, D. R., Hulley, G., Kort, E. A., Vance, N., Borchardt, J., Krings, T., Gerilowski, K., Sweeney, C., Conley, S., Bue, B. D., Aubrey, A. D., Hook, S., and Green, R. O.: Airborne methane remote measurements reveal heavytail flux distribution in Four Corners region, P. Natl. Acad. Sci. USA, 113, 9734–9739, https://doi.org/10.1073/pnas.1605617113, 2016.
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz-Sabater, J., Nicolas, J., Peubey, C., Radu, R., Schepers, D., Simmons, A., Soci, C., Abdalla, S., Abellan, X., Balsamo, G., Bechtold, P., Biavati, G., Bidlot, J., Bonavita, M., De Chiara, G., Dahlgren, P., Dee, D., Diamantakis, M., Dragani, R., Flemming, J., Forbes, R., Fuentes, M., Geer, A., Haimberger, L., Healy, S., Hogan, R. J., Hólm, E., Janisková, M., Keeley, S., Laloyaux, P., Lopez, P., Lupu, C., Radnoti, G., de Rosnay, P., Rozum, I., Vamborg, F., Villaume, S., and Thépaut, J. N.: The ERA5 global reanalysis, Q. J. Roy. Meteorol. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803, 2020.
Molod, A., Takacs, L., Suarez, M., Bacmeister, J., Song, I.-S., and Eichmann, A.: The GEOS-5 Atmospheric General Circulation Model: Mean Climate and Development from MERRA to Fortuna, Technical Report Series on Global Modeling and Data Assimilation, Tech. Rep., https://ntrs.nasa.gov/citations/20120011790 (last access: 25 July 2023), 2012.
Pandey, S., Houweling, S., Lorente, A., Borsdorff, T., Tsivlidou, M., Anthony Bloom, A., Poulter, B., Zhang, Z., and Aben, I.: Using satellite data to identify the methane emission controls of South Sudan’s wetlands, Biogeosciences, 18, 557–572, https://doi.org/10.5194/bg-18-557-2021, 2021.
Scarpelli, T. R., Jacob, D. J., Grossman, S., Lu, X., Qu, Z., Sulprizio, M. P., Zhang, Y., Reuland, F., Gordon, D., and Worden, J. R.: Updated Global Fuel Exploitation Inventory (GFEI) for methane emissions from the oil, gas, and coal sectors: Evaluation with inversions of atmospheric methane observations, Atmos. Chem. Phys., 22, 3235–3249, https://doi.org/10.5194/acp22-3235-2022, 2022.
Schuit, B. J., Maasakkers, J. D., Bijl, P., Mahapatra, G., van den Berg, A.-W., Pandey, S., Lorente, A., Borsdorff, T., Houweling, S., Varon, D. J., McKeever, J., Jervis, D., Girard, M., Irakulis-Loitxate, I., Gorroño, J., Guanter, L., Cusworth, D. H., and Aben, I.: Automated detection and monitoring of methane super-emitters using satellite data, Atmos. Chem. Phys., 23, 9071–9098, https://doi.org/10.5194/acp-23-9071-2023, 2023.
Varon, D. J., Jacob, D. J., Mckeever, J., Jervis, D., Durak, B. O. A., Xia, Y., and Huang, Y.: Quantifying methane point sources from fine-scale satellite observations of atmospheric methane plumes, Atmos. Meas. Tech., 11, 5673–5686, https://doi.org/10.5194/amt-11-5673-2018, 2018.
Veefkind, J. P., Aben, I., McMullan, K., Förster, H., de Vries, J., Otter, G., Claas, J., Eskes, H. J., de Haan, J. F., Kleipool, Q., van Weele, M., Hasekamp, O., Hoogeveen, R., Landgraf, J., Snel, R., Tol, P., Ingmann, P., Voors, R., Kruizinga, B., Vink, R., Visser, H., and Levelt, P. F.: TROPOMI on the ESA Sentinel-5 Precursor: A GMES mission for global observations of the atmospheric composition for climate, air quality and ozone layer applications, Remote Sens. Environ., 120, 70–83, https://doi.org/10.1016/j.rse.2011.09.027, 2012.
Related articles
-
CAMS: Carbon Monitor CO2 emissions documentation (Copernicus Knowledge Base)
-
CAMS: Global greenhouse gases forecasts data documentation (Copernicus Knowledge Base)
-
