The Global Registry of Fossil Fuels integrates an emissions model that converts fossil fuel production into their embedded CO2e, differentiated by country.

The emissions methodology is based on the range of constants provided by the IPCC process for combustion of fossil fuels, and for supply chain emissions on datasets from theInternational Energy Agency and Opgee for crude oil and gas, and the database of coal mines built by Global Energy Monitor.

Key documents and models exposing the detail of the emissions methodology are available on the Datasets page.

The current (0.1) version of the model uses ratios and constants established in the scientific process of the Intergovernmental Panel on Climate Change (IPCC). The latest Assessment Report (AR5, published in 2013), refers to values first published in 2006 to assess emissions from crude oil and gas.

The IPCC approach works in two stages. First, a ratio is expressed for the conversion of oil and gas into amounts of energy. Then a second set of ratios are used to convert the energy into emissions. Each of these conversion stages is expressed as a range of low, medium and high for both oil and gas which correspond to confidence intervals of P5, a weighted average in the middle, and P95.

Oil is converted from units of mass to units of energy using a range of conversion factors, ranging from the low value of 40.2 terajoules (TJ) per thousand tons to the high value of 44.8 TJ per thousand tons. For natural gas the Registry uses the volume-to-energy conversion ratio provided by the BP Statistical Analysis of 36 Petajoules per billion cubic metres.

Conversion into emissions levels is provided by a second range of constants. For oil, these vary from 71.1 (P5) through 73.3 (weighted average) to 75.5 (P95) tons of CO2e per terajoule. The ratios for gas are lower, reflecting the lower carbon intensity of gas per unit of energy produced by combustion: from 54.3 (P5) through 56.1 (weighted average) to 58.3 (P95) tons of CO₂e per terajoule.

The two separate ranges must be compounded to show the potential range of emissions per volume. For oil, one ton of crude oil could therefore produce emissions between 2.85 and 3.38 tons of CO₂e. For gas the range is from 1.95 thousands tons of CO₂e per million cubic metres to 2.10 thousand tons.

In addition, a non-fuel use (NFU) factor is added, to account for a proportion of fossil fuels that are extracted but not burned. The Registry uses the EIA 2020 NFU ratios of 12% for crude oil and 3% for gas (note: this is based on US quotas and could differ by country, but country-level estimates are not available in public domain).

Emissions from the upstream are lower than those from the combustion process of fossil fuels, but with an acknowledged wider range of variance.

For oil, the Registry uses country-level estimates provided by the Opgee research project from Stanford University. The Opgee process, which underpins California’s clean fuel standard, feeds in up to 60 variables relating to specifics of how crude oil is produced, and produces a range of estimates for emissions from aggregated country-level emissions for 91 states.

For gas, two datasets from the International Energy Agency are combined to produce country-level estimates for emissions per barrel of oil equivalent (boe) produced. The first, a data set published by the International Energy Agency in January 2021, represents non-combustion emissions from 1,155 gas projects, representing total global gas production in 2019. These are broken into various sub-components of emissions: energy used in extraction, vented CO₂, pipeline transportation, the LNG process, and methane emissions in both the upstream and downstream.

Each project has estimates and projections for a range of causes of carbon emissions in the upstream, including energy used in the extraction process, flaring, vented and fugitive methane emissions, and transportation into the refining unit. Each of the projects then has an estimate of emissions per barrel of oil equivalent attached to it.

The project data are anonymised, so individual producers cannot be directly associated with the upstream emissions intensities. But the IEA has published a second data set relating to methane emissions, which is specified at the country level. The methane projections from gas fields match the amount of methane estimated from gas fields in the first data set when factored at GWP100. This means precise projections for methane can be drawn from the second data set, leaving only the non-methane emissions to be allocated at a global level (Tier 1, according to the IPCC taxonomy of emissions methodologies).

Once the methane components are removed, the IEA data set yields a distribution of other emissions ranging from 23 kg of CO₂e per barrel of oil equivalent (P5), through a weighted average of 38 kg per boe, to a P95 estimate of 95 kg per boe. This global range is then added to the country specific point estimates for the methane components to create estimate ranges for gas by country. Although the range of the non-methane components is wide at a global level, methane itself accounts for between 60% and 80% of total CO₂e estimates from gas fields in the IEA data sets (depending on which GWP factor is used).

Coal is the largest emitter of the three fossil fuels, but generally the hardest to obtain information about.

The main issue with coal combustion is that the IPCC range estimates have margins of uncertainty that are much wider than for oil and gas. Whereas the maximum variation in IPCC ranges for crude oil is 18%, and for gas only 8%, for anthracite coal, for example, it is 60%, and 400% for lignite.

The Registry has incorporated the granular database of project partners Global Energy Monitor. This contains production information and sub-types of coal for over 2,300 coal projects around the world, totaling over 90% of coal production according to various sources that estimate production at national level, such as the BP Statistical Analysis and data from the US Energy Information Administration.

To produce country-level estimates of emissions, Monte Carlo stochastic modeling is used on the Coal Tracker Database. The production and sub-type of coal for each mine is listed, and then, from the nine possibilities under the IPCC Tier 1 projections for each sub-type of coal, one is chosen at random. This gives a total estimate of emissions from all the mines in the country. Then the process is iterated 5,000 times for each coal producing country.

Then the totals of each of the 5,000 calculations are lined up from smallest to largest and confidence intervals (P5, weighted average, and P95) are selected. The gap between the sum of production of the mines in the database in any given country and the total estimated production of that country is filled in pro rata. The resulting ranges are an order of magnitude smaller than using a pure Tier I approach. They are still considerable in some countries – India shows a 74% range and Germany 72%, owing to the predominance of lignite production. But average variance by country is 25% - still higher than oil and gas, but producing manageable margins of uncertainty.

The Registry does not include an estimate for all operational CO₂e emissions from coal mines, just methane. It does this by using the fields in GEM’s Coal Tracker Database which relate to the depth and sub-type of mine. This allows deployment of an estimations method developed by Ray Pilcher and others on Langmuir equations , which estimate the amount of methane trapped in coal deposits at various depths under the surface, and released when it is mined.

For mines where GEM’s Coal Tracker indicates a depth as exact, a point calculation is made. Where the depth is approximate, a margin of uncertainty 10% up and down is applied.

The Registry does not currently estimate non-methane CO₂e emissions from working mines, or methane emissions from abandoned mines.

The Registry considers it important to include emissions estimates for both methane factorisations to CO₂e currently within the IPCC process, GWP100 and GWP20.

GWP20 gives a much higher factorisation of methane to carbon dioxide: in the IPCC's Fifth Assessment Report, GWP100 was given as a range between 28 and 34 compared to CO₂, whereas GWP20 lies in a range between 84 and 87. The reason is that methane is a much more potent greenhouse gas than carbondioxide but also breaks down in the atmosphere, leading to it becoming less potent over time.

For coal and gas, the Registry has point emissions estimates for methane at country level. These can therefore be directly be converted from one Global Warming Potential factor to the other by swapping the multiple applied to CO₂e.

Because the Opgee country-level upstream emissions for crude oil do not provide disaggregated estimates for methane emissions at either factor, an indirect method has to be used.

The IEA publishes a similar upstream emissions data set for 2,055 crude oil projects, representing global production, as for gas. It is broken down into constituent elements, including methane factored at GWP100. The IEA crude oil dataset can therefore be refactored from GWP20 to GWP100, with an overall difference of 64% across the entire global set. This factor is therefore applied to the Opgee total upstream estimates (using GWP100) to derive an Opgee-based GWP20 estimate for crude oil emissions by country.

Although GWP100 has been deployed more often to build the estimates included in Greenhouse Gas inventories, the IPCC process states, when offering both ratios, that it does not recommend one rather than another. GWP20 could be of greater analytical value in contexts relating to study of potential feedback loops, for example. The state of New York adopted GWP20 as its main factorisation in regulations in 2019, as does the International Methane Observatory launched by the United Nations Environmental Program, launched in 2021.

It should also be noted that debates about how to factorise methane to create carbon dioxide equivalence figures are ongoing. As well as a debate over the choice of GWP100 versus GWP20, some experts also suggest the GWP approach should be replaced by Global Temperature Potential (GTP), which was introduced into the IPCC process in 2005.. Some researchers are also suggesting adaptation of the existing GWP methodology to so-called "GWP*" (GWP Star) calculations..