Electricity carbon footprint

← Analyses

How much greenhouse gas is associated with one kilowatt-hour of electricity? The answer depends on the source, the system boundary and the method. This benchmark brings monetary input–output estimates and physical-unit sources onto a common basis, g CO₂eq/kWh.

The comparison follows Citterio et al. (2026), Carbon intensity of electricity: a systematic methodological and quantitative review, and the accompanying eNextGen analysis. The charts use the blogpost snapshot, which includes updates beyond the article, including nxsut 3.0.

By country and data source

Each marker is an estimate for one country from one source. Colour distinguishes natively monetary and natively physical factors; marker shape identifies the source. Change the year and scope, select countries, or use the legend to hide sources. The table follows the active filters. Downloaded figures retain their source credit. For the published comparison data and their source-specific reuse terms, follow the article and repository links below; this page does not export an emission-factor dataset.

Carbon footprint and intensity of electricity, by country and data source

Each marker is one data source's estimate for a country; colour = natively monetary (converted to a physical basis via electricity prices) vs natively physical, shape = source. Filter by scope, year and methodology. Differences reflect source data, system boundaries and methodological choices.

Scope Direct Life-cycle

Year 2017 2023

Method IO LCA Other

Chart Table

Full screen JPG

Swipe across the chart to see the full range, or switch to Table.

^* Natively monetary factors are converted to a physical basis (gCO₂eq/kWh) using country-level electricity prices. Snapshot of the eNextGen blogpost, generated from nxbase. No live data query is made by this page. For Eora, GLORIA and Electricity Maps, only the datapoints shown in the article are held in nxbase. Source: Citterio et al. (2026), doi:10.1007/s44498-026-00148-3.

Assumptions behind the country comparison

Direct and life-cycle

Direct intensity counts emissions associated with electricity generation. Life-cycle footprints also trace upstream supply-chain contributions within each model's boundary. Different sources do not necessarily include the same processes, capital goods or transmission and distribution losses.

Monetary to physical

A monetary factor is multiplied by a country-year electricity price: g CO₂eq/EUR × EUR/kWh = g CO₂eq/kWh. Currency scales and exchange rates must be consistent. The snapshot pipeline uses country-year prices and ECB exchange rates where required. The chosen price is part of the estimate, rather than a universal unit conversion.

Generation and electricity delivered

Production-based factors and consumption-based factors answer different questions. Electricity imports, exports and losses can change the delivered mix; matching units does not remove these differences in system boundary. nxsut's national comparison uses its electricity-need commodity.

Years and missing observations

The year filter selects the comparison year represented in the snapshot. The underlying source release and reference-factor vintage may differ. Only available observations are shown; an absent point does not mean zero. The country sample contains 24 EU countries and the United States; Cyprus, Luxembourg and Malta are omitted, following the country comparison.

GHG characterisation

The nxsut results use GWP100 characterisation with IPCC AR5 factors in the snapshot pipeline. Other sources retain the characterisation and gas coverage of their published estimates. A common CO₂-equivalent unit does not guarantee identical gas coverage or characterisation.

By generation technology

National averages conceal the differences between generation technologies. These distributions pool country × technology estimates from nxsut 3.0, EXIOBASE Hybrid and BONSAI, together with the article’s literature references. Select sources to see how each contributes to the distribution.

BONSAI uses its native b2 HIOT, with reference year 2016 and inputs updated on 21 October 2026 (Europe/Rome). Its direct intensities and life-cycle footprints use the native electricity-generation products, with one observation per available individual country and technology. The date is shown beside the source controls and retained in downloaded figures.

Carbon footprint of electricity, by generation technology

Physical-unit emission factors (gCO₂eq/kWh), pooled across country × technology estimates and literature references. Fossil fuel categories and other technologies. Each box spans the interquartile range (median = solid line, mean = dashed); whiskers reach 1.5×IQR; dots are outliers. Switch between direct and life-cycle emissions, and zoom on the other technologies.

Scope Direct Life-cycle

Show All Other technologies only

Chart Table

Full screen JPG

Fossil Other technologies IQR + median mean outlier

Swipe across the chart to see the full range, or switch to Table.

Physical-unit emission factors by technology, gCO₂eq/kWh, from the blogpost sample. The input–output sources nxsut 3.0, EXIOBASE Hybrid and BONSAI contribute one point per available country × technology. The literature (Electricity Maps, IPCC, NREL, Ecoinvent) is the article's supplementary distribution: Citterio et al. (2026), doi:10.1007/s44498-026-00148-3. Life-cycle axis capped at 2500 gCO₂eq/kWh (outliers beyond drawn at the edge).

Assumptions behind the technology distributions

What the boxes represent

Each box summarises the observations currently selected: the middle 50% of values, median, mean and whiskers up to 1.5 times the interquartile range. Sources contribute different numbers of observations. The sample is not weighted by electricity generation or balanced to give every source equal weight, and the boxes are not statistical confidence intervals.

Physical output

The model estimates use activity-side direct intensity and life-cycle footprint, converted from t CO₂eq/TJ to g CO₂eq/kWh. A national average electricity price is not used to convert individual generation technologies: their specific prices may differ substantially.

Technology groups and biomass

Technology labels follow the blogpost aggregation. “Other technologies” includes nuclear, hydro, wind, solar, geothermal, tide and wave, and biomass and waste. This grouping does not imply that every included technology has a low or zero footprint. The model characterisation gives biogenic CO₂ a factor of zero; other emissions and upstream contributions can remain.

BONSAI

Native C_elec_POW* products are paired with A_POW* generation activities and grouped using the supplied classification. POWP is petroleum (Oil), POWO is tide/wave/ocean, POWM is geothermal and POWS is photovoltaic solar. Direct intensity uses characterised E divided by source reference production; the life-cycle footprint uses MARIO's e(I-z)^-1. Native t CO₂eq/TJ are multiplied by 3.6, without a price conversion. For this explorer, native gas rows are recharacterised with AR5 GWP100: fossil CO₂ 1, both methane origins 28 and N₂O 265, following the EDGAR convention used elsewhere in the explorer. Contributions of iLUC to these gases are retained. Accelerated CO₂ is excluded because no AR5 factor has been verified; F-gases are not characterised. This three-gas subtotal differs from the official BONSAI AR6 indicator and MARIO's BONSAI default. RoW and Global blocks are excluded from this country sample; missing country/technology nodes are not filled.

Sample and display limits

The chart uses the pooled blogpost sample, rather than a common country-year panel across every source. Values outside the plotted range are drawn at the edge. The table reports uncapped statistics; the CSV retains the individual observations. Categories with fewer than three observations appear as individual points.

Sources and reproducibility

The input–output calculations are performed with MARIO; nxbase brings the resulting indicators and external reference factors into a common structure. The charts on this page are fixed snapshots of that analysis. They work without an account or a live API connection.

BONSAI’s technology observations were calculated separately with MARIO parse_bonsai(table="IOT") from frozen inputs, added on 21 October 2026. On identical inputs, MARIO and BONSAI’s dimensional formulation agree over all 53,750 production nodes (official AR6 GWP100 relative L2 difference 5.83e-16), with no partial solve. This verifies calculation equivalence; it does not compare against an older saved footprint CSV. The explorer’s AR5 values are a linear combination of the verified native gas multipliers; the original AR6 audit is retained separately. The BONSAI technology snapshot records source timestamps and hashes, the explicit technology mapping, characterisation, native country/product identities and verification. See the BONSAI walkthrough for parsing and calculating its footprints. The source folder label 20260615 is retained as provenance; its contents are updated in place.

Read the article ↗ Read the eNextGen analysis ↗ Code and data ↗

Charts and the original author-derived comparison data are adapted from the eNextGen analysis and its accompanying repository, with attribution under CC BY 4.0. Third-party observations remain subject to their source terms; this page includes the comparison snapshot, rather than complete source databases. The additional BONSAI technology results are attributed to BONSAI and calculated with MARIO; they are not part of the article or eNextGen snapshot. See the source repository and blogpost for source-specific licences and redistribution limits.