methodology

Power sector emissions are estimated by first assembling a global geolocated inventory of power plants, generation, and metered emissions data. Machine learning models then predict plant generation from satellite images. These predictions are aggregated and combined with available generation data and carbon intensity factors to derive emissions estimates.

Full Methodology

methodology

Data from EDGAR calculated after subtracting energy industry subsectors directly measured by Climate TRACE. As Climate TRACE capabilities expand, this “other” category will shrink. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Forest clearing emissions are calculated by combining satellite-derived estimates of aboveground biomass from forested areas in the year 2000 and tree cover loss by year. These estimates are interpreted as “committed emissions,” meaning that the mechanism and timing of atmospheric carbon emissions are unknown. Our methods are based on work by Zarin et al (2016) and the Global Forest Watch project. For more information and documentation, visit https://github.com/carbonplan/trace.

Full Methodology

methodology

Forest-fire emissions are estimated with an empirical relationship using inputs from satellite-derived datasets and vegetation-specific combustion and emissions factors. Forest-fires are identified within a country, emissions estimated, and aggregated to the country-level for total GHG emissions.

Full Methodology

methodology

Savanna-fire emissions are estimated with an empirical relationship using inputs from satellite-derived datasets and vegetation-specific combustion and emissions factors. Savanna-fires are identified within a country, emissions estimated, and aggregated to the country-level for total GHG emissions.

Full Methodology

methodology

Shrubland-fire emissions are estimated with an empirical relationship using inputs from satellite-derived datasets and vegetation-specific combustion and emissions factors. Shrubland-fires are identified within a country, emissions estimated, and aggregated to the country-level for total GHG emissions.

Full Methodology

methodology

Aluminum emissions are estimated with a disaggregation method to determine a plant’s share of national capacity. Relevant emissions factors for the plant are applied to derive emissions for a given year.

Full Methodology

methodology

Cement emissions are estimated with a combination of modeling with satellite imagery and plant-level capacity to generate asset-level cement production. The asset-level cement production is calibrated against country-level production. Finally, plant-level emission factors are applied to derive total emissions.

Full Methodology

methodology

Data from EDGAR derived after subtracting manufacturing subsectors directly measured covered by Climate TRACE. As Climate TRACE capabilities expand, this “other” category will shrink. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Ammonia emissions are estimated with a disaggregation method to determine a plant’s share of national capacity. Relevant emissions factors for the plant are applied to derive the emissions for a given year. Soda ash manufacturing emissions are derived from utilisation rates obtained from country level data, split by natural and synthetic soda ash producing countries, and projected to plant level to proxy for plant level production. Methanol manufacturing emissions are derived from average utilisation rates by country projected to the plant level. Where data were unavailable, a relationship between country-level installed capacity and utilisation rates was used to fill in gaps.

Full Methodology

methodology

Pulp manufacturing emissions are estimated with a disaggregation method to determine a plant’s share of national capacity. Relevant emissions factors for the plant are applied to derive the emissions for a given year. For paper manufacturing emissions, utilisation rates obtained from country level data are split by paper type and projected to plant level to proxy for plant level production.

Full Methodology

methodology

Steel emission estimates use two subgroups. 1) (BF), basic oxygenation furnace (BOF) and open hearth furnace (OHF) plants are modeled with satellite imagery and plant-level capacity to generate asset-level production calibrated against country-level production. 2) electric furnace (EAF) plants use a disaggregation model to compute its national capacity share.

Full Methodology

methodology

Historical flight status data for all passenger and freight flights is combined with the International Civil Aviation Organisation (ICAO)’s methodology to estimate aviation emissions globally. Estimates available from 2018 to 2020. 2018 values have been back-filled for the remaining years.

Full Methodology

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Road transportation emissions estimates for 20 countries are generated by running machine learning models on satellite imagery and road and building masks, and population data. These models are trained on annual bottom-up conterminous U.S. CO2 inventories. Estimates available from 2017-2019 and back/forward-filled for the remaining years for consistency. EDGAR data compiled for all remaining countries. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

Full Methodology

methodology

Estimation of crop fire emissions uses an empirical relationship with inputs from satellite-derived datasets and vegetation-specific combustion and emissions factors. Crop residue fires are identified within a country, emissions estimated, and aggregated to the country-level for total GHG emissions.

Full Methodology

methodology

Data compiled from FAOSTAT under the source category “Enteric fermentation”. Estimates are available from 2015 to 2019. 2019 values have been forward-filled for 2020.

methodology

Synthetic fertilizer N2O emissions for 16 countries estimated with a semi-empirical model that uses country-level crop production and yields; FAOSTAT data compiled for all remaining countries. Estimates available from 2015 to 2019. 2019 values have been forward-filled for 2020.

Full Methodology

methodology

Data compiled from FAOSTAT from the source category “Manure Management”. Estimates are available from 2015 to 2019. 2019 values have been forward-filled for 2020.

methodology

Methane emissions for the year 2020 for 23 countries estimated by identifying rice growing areas using satellite-based mapping. For prior years and all other countries, FAOSTAT data has been used. Data back or forward-filled to ensure consistency.

Full Methodology

methodology

Oil and gas field- and basin-level carbon dioxide and methane emissions are estimated with the Oil Production Greenhouse Gas Emissions Estimator (OPGEE) Model. Inputs include a number of oil and gas field characteristics, including satellite-derived flaring volumes. Field-level estimates are summed and scaled to country-level on an annual basis.

Full Methodology

methodology

Oil refining emissions are estimated with the Petroleum Refinery Life-Cycle Inventory Model (PRELIM). PRELIM estimates energy use and GHG emissions associated with processing various crude oil types for different refinery configurations. Inputs include refinery locations, configurations, and crude refining throughputs to generate GHG intensities and absolute emissions by country.

Full Methodology

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Data not yet available.

methodology

Data available for Annex 1 countries from UN Data.

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Data from EDGAR. Estimates available from 2015 to 2018. 2018 values have been forward-filled for the remaining years.

methodology

Shipping estimates are estimated using Automatic Identification Systems (AIS) transmissions from vessels, which is combined with vessel characteristics. The estimation method is based on audited emissions reported to the European Union. Values for 2015-2017 were backfilled from 2018 estimates and adjusted based on the overall trend in emissions reported in the 4th IMO Report. Values for all years were adjusted to compensate for vessels not included in the model.

Full Methodology

methodology

Bauxite-mining emissions are derived from total production at yearly, country-level scale using a minimum of 10 years of production data (2007-2016) and forecasted through 2020. Emissions factors are applied based on the ore quality within a country and the energy required to extract based on surface proximity.

Full Methodology

methodology

Data included in ‘Oil and Gas - Solid fuel transformation’.

methodology

Copper mining emissions are derived from total production at yearly, country-level scale using a minimum of 10 years of production data (2007-2016) and forecasted through 2020. Emissions factors are applied based on the ore quality within a country and the energy required to extract based on surface proximity.

Full Methodology

methodology

Iron mining emissions are derived from total production at yearly, country-level scale using a minimum of 10 years of production data (2007-2016) and forecasted through 2020. Emissions factors are applied based on the ore quality within a country and the energy required to extract based on surface proximity.

Full Methodology

methodology

Rock-quarry emissions are derived from total production inventories at yearly, country-level scale using a minimum of 10 years of production data (2007-2016) and forecasted through 2020. Then emissions factors are applied based on the ore quality within a country and the energy required to extract based on surface proximity.

Full Methodology

methodology

Sand-quarry emissions are derived from total production inventories at yearly, country-level scale using a minimum of 10 years of production data (2007-2016) and forecasted through 2020. Then emissions factors are applied based on the ore quality within a country and the energy required to extract based on surface proximity.

Full Methodology