Skip to main content

You are viewing a project that is currently in draft state for the Standards Working Group in the Green Software Foundation. This project should not be considered finished or officially supported in any way by the Green Software Foundation or its members.

Carbon Intensity and Data Sources

In short: every operational component needs a carbon intensity: how much CO2eq the grid emitted for each kWh. Use location-based figures for the places where the electricity was actually used, say where they came from, and say how fine-grained they are. This page lists sources the community already uses.

(Named providers, models and tools on this page last reviewed: 2026-10. The specification names none of them; inclusion here is not an endorsement.)

What the specification asks​

  • Location-based only. Clause 7.1 requires location-based carbon intensity for every operational component, and Clause 11 excludes market-based measures. Renewable energy certificates, power purchase agreements and "green hosting" claims based only on such purchases do not change the figure you use.
  • Disclosed. Clause 8 item 7 requires the carbon intensity sources, and their spatial and temporal granularity, to be disclosed.
  • Per region. Clause 7.2 requires server carbon intensity to be applied per grid region; Clause 7.4 asks for client intensity to reflect where users are, where that is known.

Grid carbon intensity sources​

SourceWhat it offers
Electricity MapsLocation-based grid carbon intensity by zone, historical and real-time
WattTimeGrid emissions data by region, historical and real-time
International Energy Agency (IEA)National emission factors
Regional and national grid operatorsPublished carbon intensity for their grid
UNFCCCNational emission factors

Choose the most granular source that covers your regions, and use the same source for the baseline and the comparison when evaluating an action (Clause 9).

Temporal granularity​

The original draft linked granularity to maturity: annual averages at Entry, daily at Standard and hourly at Advanced. The table is on Implementation Tiers and Data Quality. Finer granularity lets the score reflect when work happens, which matters for carbon-aware scheduling.

Models and reference datasets​

The original draft's bibliography named these models and datasets. They are referenced from the component pages where they apply:

  • Sustainable Web Design Model (SWDM) v4, Sustainable Web Design Community Group: sustainablewebdesign.org. A data-transfer model; used for O_network only as the last-resort method in Clause 7.3. See Network Emissions.
  • CO2.js, Green Web Foundation: developers.thegreenwebfoundation.org. A JavaScript library implementing data-transfer models, including SWDM.
  • Boavizta environmental footprint data: boavizta.org. Embodied emissions for servers, cloud instances, devices and network hardware.
  • Cloud Carbon Footprint methodology: cloudcarbonfootprint.org. Coefficients for estimating cloud energy and embodied emissions.
  • GHG Protocol ICT Sector Guidance: ghgprotocol.org. Sector guidance on allocating ICT emissions, useful background for allocation choices.

A component-by-component list of tools is on the Tools Directory.

Please submit any comments you have here.