Global Methane Emitters Tracker methodology

This article is part of the Global Methane Emitters Tracker, a project of Global Energy Monitor.

Download full dataset

Report an error

Related categories:

Coal mines Oil and gas extraction Oil industry pipelines LNG Terminals

Global Energy Monitor publishes the Global Methane Emitters Tracker (GMET), a dataset that provides estimates of fossil fuel emissions at oil and gas and coal extraction sites, natural gas transmission pipelines, proposed fossil fuel projects and reserves, and attribution of remotely-sensed methane plumes.

As of November 2023, the first version of the tracker includes methane emissions estimates for coal extraction and gas pipelines, and attributions of remotely-sensed methane plume observations for oil and gas infrastructures in North America, and coal mine observations worldwide. GMET also associates assets within GEM’s Oil & Gas Extraction Tracker to the methane emissions estimates developed by Climate TRACE.

The tracker plans to expand its remotely-sensed plume attribution coverage in future iterations.

Methodology

Global Methane Emitters Tracker uses asset level infrastructure data in Global Energy Monitor’s Global Coal Mine Tracker, Global Oil and Gas Extraction Tracker, and Global Gas Infrastructure Tracker and other non-GEM datasets for attributions and emissions.

Plume Attribution

1.1 Coverage

As of November 2023, Global Energy Monitor has analyzed 1,243 plumes created by CarbonMapper plane flights using the Global Airborne Observatory (GAO) and the Airborne Visible Infrared Imaging Spectrometer - Next Generation (AVIRS-NG) in five states in the United States. Each of these plumes is a snapshot of methane emissions taken a particular place and time (for more details on how the imagery is made, see CarbonMapper.) In addition, we compared global plumes created by the NASA Earth Surface Mineral Dust Source Investigation (EMIT) satellite to GEM’s Global Coal Mine Tracker (GCMT), attributing 17 plumes to specific coal mines.

For this first release of the tracker, Global Energy Monitor has focused on these plumes because (to the best of our knowledge) these observations have not been systematically and publicly attributed at the wellpad or field scale. Additionally, we focus on states for which we have field-level data within GEM’s Global Oil and Gas Extraction tracker. Note that the Pennsylvania plumes analyzed here are all outside the Marcellus shale, as we have attributed Marcellus plumes using a different protocol and intend to roll them into the next update of the Global Methane Emitters Tracker (GMET). California plume images created by CarbonMapper in California between 2016-2018 have been analyzed by Rafiq et al. (2020), and CarbonMapper plumes in the Permian Basin have been attributed by the Environmental Defense Fund’s PermianMAP project.

1.2 Attribution Methodology

All of the plume attributions in GMET were done manually following a protocol similar to Rafiq et al. (2020). This involves visually inspecting points representing the origin of each plume observation, along with raster images of the methane detections to both high-resolution Google Earth basemap satellite imagery and a variety of infrastructure datasets enumerated below.

The RGB images associated with the plumes created through plane flights were compared to the Google Earth imagery, to ensure that the infrastructure used for attribution existed at the time of the emission. The attribution of non-US plumes to coal mines (which relied on satellite-based imagery with no available corresponding RGB image) were verified by ensuring that the coal mines in GEM’s database were all present at the time of the plumes’ emissions. Plume origin points were analyzed both individually and in context with other plumes which CarbonMapper had determined were interconnected (i.e., the points represented multiple observations of the same infrastructural source.) Each plume observation was reviewed by at least two GEM analysts.

GEM provides attribution data in the following fields within GMET: Nearest Government Well ID, GEM Infrastructure Integration, Type of Infrastructure, California Vista Nearby Assets, and Notes. Note that because oil and gas infrastructure can change hands regularly, that the operator assigned to the asset in each of the databases presented below may not match the operator at the time of the plume’s emissions.

• Nearest Government Well ID: In every state except Pennsylvania, this field provides an American Petroleum Institute (API) number. We use API Number here as a stand-in for the whole wellpad–not necessarily the well itself. Plumes are assigned a value in this field if the plume origin is within 10m of a well with no other obvious sources of methane, or if the plume is within a clearly defined wellpad. Because these IDs are used to represent the entire wellpad, it is possible that the actual source of the emission is a storage tank, pipeline leading from the well, or another well on the wellpad. For multi-well pads, the ID of the closest well with government data is provided.

• Type of Infrastructure: These categories of infrastructure are drawn largely from the definitions used by California’s Statewide methane emissions inventory, Vista-CA, definitions, with a few changes. These changes include: 1) wellpad in GMET refers to either an oil or gas well or infrastructure (storage tank, flare stack, etc.) located within a clearly defined wellpad 2) compressor stations are identified for assets not contained in the Vista-CA data 3) offshore platforms are identified 4) dairies and livestock facilities are combined.

• GEM Infrastructure Integration: This field contains the name of the GEM asset in which the plume origin falls. For assets identified through GEM’s Global Coal Mine Tracker, Global Oil and Gas Plant Tracker (GCPT), and Global Gas Infrastructure Tracker (GGIT), plumes are attributed if they are located within the asset contained within these databases. Note that for some California plumes, GEM assets may not be referenced if they have been attributed to a Vista-CA asset. For the Global Oil and Gas Extraction Tracker (GOGET), plumes which have been attributed to a particular wellpad have been associated to the appropriate GOGET oil and gas extraction area containing the well. Note that the ownership data within GOGET may be more or less recent than the emissions date of the plume, i.e., the company listed as operator for a GOGET asset containing the plume may not have been the operator at the time of emission.

• Type of Infrastructure: Where possible, the source of a plume is assigned to a broad infrastructure category. These categories are drawn largely from the Vista-CA definitions, with a few changes: 1) wellpad here refers to either an oil or gas well or infrastructure (storage tank, flare stack, etc.) located within a clearly defined wellpad 2) compressor stations are identified for assets not contained in the Vista data 3) offshore platforms are identified 4) dairies and livestock facility categories are combined. Plume observations are assigned an infrastructure type based on a combination of visual inspection, designations within Vista-CA, and comparisons against GEM’s databases and the government sources enumerated below. In rare cases, Google Maps data are used to identify an infrastructure type. In these instances, the Google Maps data are verified against either a government source or company reports identifying the asset.

• California Vista Nearby Assets: California plumes whose origins fall within a Vista-CA asset (for the infrastructure types represented by polygons) or who are located on the same facility identified in Vista-CA (for infrastructures represented by points) are assigned the ID for the asset within Vista-CA. For Vista-CA oil and gas fields, note that in rare cases the plume, even if it falls within a Vista-CA field, may not be related to oil and gas production. Refer to the qualitative notes for more details in these instances.

• Notes: GEM researchers made qualitative observations of the plume’s location and the surrounding infrastructure. Each set of notes was reviewed by at least two GEM analysts. For instances where attribution is obvious (e.g., the plume falls directly over an Oil & Gas Facility identified in the Vista-CA data) notes are not taken. Broadly, the notes field offers the justification GEM researchers used when making attributions. They are particularly useful for identifying instances where definite attributions were challenging: e.g., where there was no visible infrastructure in the basemap and a possible underground pipeline may have been responsible, or when a plume observation is between multiple pumpjacks or wellheads. Additionally, they can provide more detail on the specific equipment from which a plume may originate, such as a flarestack on a wellpad. Notes are reviewed to ensure that there is consistency among plumes believed to be originating from a single source (though the exact wording may vary between plume observations).

1.3 Data sources used for attribution

1.3.1 Global Energy Monitor (GEM) Databases

Plume locations were compared to assets within GEM’s Global Coal Mine Tracker, Global Oil and Gas Plant Tracker, Global Gas Infrastructure Tracker, and Global Oil and Gas Extraction Tracker.

The infrastructure data on coal mine, oil and gas units, and pipelines are collected from and validated through five main sources:

• Government data on individual units, country energy and resource plans, and government websites tracking extraction permits and applications.
• Reports by state-owned and private companies;
• News and media reports;
• Local non-governmental organizations tracking extraction permits and operations;
• On-the-ground contacts who can provide first-hand information about a project.

1.3.2 Non-GEM Databases by State

Below is a list of all of the non-GEM databases we used for attribution in each state. In all states, for pipelines we relied on observations of infrastructure visible in the Google Earth imagery. Plumes which are above areas with no visible infrastructure and may be connected to an underground pipeline are described as such in the notes column.

• California: The attribution data we drew on for California included all publicly available information contained within the 2019-12-17 Vista-CA dataset. This includes power plants, refineries, natural gas fueling stations, natural gas stations, oil and gas fields, processing plants, storage fields, feed lots, digesters, dairies, landfills, composting sites, wastewater treatment plants, and oil and gas wells (though for these we use a more recent version of the underlying California Geological Energy Management (CalGEM) well data, downloaded 6/15/2023). Ohio. We downloaded oil and gas well and wellpad data from the Ohio Department of Natural Resources on 9/12/2023.
• Louisiana and the Western Gulf: Louisiana Oil and Gas wells were downloaded from the Lousiana Departmnent of Natural Resources Strategic Online Natural Resources Information System (SONRIS) on 9/20/2023. Federal offshore platforms and lease polygons were downloaded from the Bureau of Ocean Management (BOEM) on 9/12/2023.
• Pennsylvania: Pennsylvania well data were downloaded from Pennsylvania Department of Environmental Protection on 5/24/2023.
• Texas: Texas wells were downloaded from the Texas Railroad Commission on 8/27/2023.

Oil and Gas Extraction Areas - connecting to ClimateTRACE’s estimations

These oil and gas fields are drawn from GEM’s Global Oil and Gas Extraction Tracker. For methane analysis purposes, we have added three fields in the data in order to relate these assets to the oil and gas extraction areas within ClimateTRACE’s database, which itself is built on the Rocky Mountain Institute’s Oil Climate Index plus Gas (OCI+). Where the GOGET assets differ from the TRACE fields, they are typically more granularly defined, particularly for unconventional fields in the United States and Canada. In other cases, the TRACE assets are defined based on fuel type (i.e., Eagle Ford - Dry Gas and Eagle Ford - Volatile Oil), while the GOGET assets in the same region are not separated out in this fashion. The emissions and emissions factor columns in the GEM Global Methane Emitters Tracker data therefore do not correspond directly with the GOGET assets, but rather the TRACE assets in which the GOGET are contained.

The granularity of the GOGET assets is part of why TRACE field emissions measurements were appended here, rather than applying an emissions factor to GOGET fields themselves. Official emissions factors, such as those from the EPA Greenhouse Gas Emissions Inventory, tend to undercount methane emissions from oil and gas production by a roughly two-fold, and they become less reliable at smaller scales owing to the stochastic behavior of super-emitters (See more in the literature underlying the OCI+/TRACE model here Rutherford et al., 2021). Detailed component and equipment counts within each GOGET field would be necessary to perform simulations such as those undergirding the OCI+/TRACE emissions estimates.

Estimating methane emissions from transmission pipelines

The gas pipelines analyzed in GMET are drawn from GEM’s Global Gas Infrastructure Tracker (GGIT). In order to estimate emissions, we use a Tier 1 estimation drawn from the 2019 Refinement to the IPCC (Table 4.2.4i, Tier 1 Emission Factors for Gas Transmission and Storage segment.) Because GEM does not yet have compiled comprehensive data on leak detection and repair (LDAR) across pipeline operators, we use the default emissions factor of 4.1 (-20%/+30%) tonnes per kilometer. For pipelines with robust LDAR systems this will be an overestimate, as the emissions factor for pipelines which have extensive LDAR and >50% dry seals for centrifugal compressors have an IPCC Tier 1 emissions factor of 2.8 (-20%/+30%). Therefore, we have explicitly named this column in the data as the emissions for pipelines assuming no LDAR, if operational. We include estimates for non-operational pipelines so that users can estimate the potential emissions of the pipelines if they were to be used. For details on how pipeline lengths are estimated, refer to the GGIT methodology.

Estimating potential methane emissions from oil and gas field reserves

Broadly, these are back-of-the-envelope estimates created by multiplying the quantity of Proved & Probable (2P) reserves within each GOGET asset against a regionally specific emissions factor. The emissions factors used in this estimation are based on an extrapolation of the Oil Climate Index Plus Gas (OCI+) and are specific to fuel type (either oil, gas, or condensate). For GOGET assets reporting quantities in other fuel categories (e.g. “hydrocarbons”) estimates are not developed. Additionally, the emissions factor chosen here represents just Scope 1 emissions directly related to oil and gas production (i.e., not indirect emissions related to energy consumption, end-consumer use, etc.) Therefore, the values produced here are best interpreted along the lines of: “If these entire reserves were extracted under current operating procedures, what would the methane emissions be from their production?” This line of reasoning is similar to the one developed by Heede & Oreskes (2016) though here we focus only on methane, and include Probable reserves.

The biggest challenge for creating these estimates are the differences in definition of what constitutes an oil and gas “field” (i.e., extraction area) between GEM’s GOGET dataset, and OCI+/ClimateTRACE, which relies on commonly-used but proprietary Rystad field data. GOGET assets which have been associated with a TRACE field are assigned the upstream methane emissions factor for that field, for the latest data year available within the OCI+ dataset, using a 100-year global warming potential (GWP). (20- and 100- year GWPs for this emission factor are no more than 1% different from one another.) GOGET assets which are not associated with a TRACE asset are assigned the average emissions factor for its country within the OCI+ data.

There are important limitations to using a sophisticated modeling framework like OCI+--which builds off of component- and equipment-level data of oil and gas facilities and is validated against remotely sensed methane data—and extrapolating it to a back-of-the-envelope estimate like the one provided here. For one, country-level processes may not drive differences in emissions between oil and gas extraction areas, particularly if there is significant in-country geological variation between regions, differences in the number and proportion of components (valves, thief hatches, etc.), and work processes (liquid unloadings, workovers, etc.). Additionally, methane emissions in oil and gas production are driven in large part by super-emitters originating in accidental releases (see Alvarez et al., 2018). Especially at scales as small as the field level, these stochastic processes are unlikely to remain static year-to-year, particularly under changing leak detection and repair regimes. Because of these limitations, the estimates presented here should be construed as exploratory.

Coal mining methane emissions

On coal mining emissions, GEM has relied on its methodology developed in 2022 through the Global Coal Mine Tracker.