globalwarmingindex.org

Global Warming Index - Tracking progress to a safe climate

Methodology

The Global Warming Index attribution method estimates contributions to observed warming by separating out the contributions from various human-induced and natural forcing drivers of climate change, and natural year-to-year fluctuations known as internal variability.

The tree diagram below depicts the hierarchy separation of observed global mean surface temperature into its constituent components:

The attribution method separates the warming contributions through a series of simple steps (labelled correponding to the animation below):

Step 1: Aggregate emissions and effective radiative forcings data: Collect latest updated timeseries for all climate drivers (forcings), including all natural and anthropogenic sources [IGCC 2025, following IPCC AR6 WGI Ch.7 method] . The individual forcings are aggregated into three main components: natural forcings, greenhouse gas forcings, and other human forcings (see above).

Step 2: Estimate temperature response using simple model: A simple climate model that is commonly used in the IPCC [FaIR v2.0 (Leach et al., 2021)] is used to convert the forcing timeseries into initial estimates of the temperature response for each component. Warming is defined relative to the standard 1850-1900 baseline, but but can be simply re-calculated for any baseline as needed.

Step 3 & 4: Combine components and compare to observations: Sums are performed to combine the different warming contributions to get the total forced warming estimates: human-induced, and total forced warming. The estimated total forced warming is compared to observed global mean surface temperature data, with the difference between the two defined as the residual.

Step 5: Constrain forced warming estimates: A simple least-squares fit is used to constrain the estimated forced warming components to best match observations, and account for alternative possible realisations of internal variability. These additional constraints allow us to scale each of the estimated forced warming components, improving both the accuracy and precision for each component. This is why the uncertainty ranges narrow significantly between steps 4 and 5 below. This optimisation also effectively removes any signal from forced warming from the residual which therefore represents the remaining internal variability in the observations that cannot be explained by the forced response.

Step 6: Extract Global Warming Index: While the full breakdown of contributions to observed warming is available via this method, the Global Warming Index focuses on the human-induced warming component as the key metric of interest for tracking progress towards international climate goals.

Uncertainty: Uncertainty is estimated by repeating the above calculations for a very large Monte Carlo ensemble of possible variations of: (i) forcings ensemble (1000 members of ERF ), (ii) observational data (200 members of HadcRUT5 ), (iii) climate response parameters (~30 members of 6 FaIR emulation parameters of CMIP6 simluations) and (iv) alternative internal variability realisations (200 members of CMIP6 pre-industrial control simulations). Calculating any one of the above steps for a single member of this ensemble is computationally inexpensive, but the full uncertainty analysis requires repeating the full calculation for all combinations for an extremely large Monta Carlo ensemble containing hundreds of millions of members, and therefore is only possible through the use of high-performance computing clusters.

Generalisation The simplicity of the method, and use of the FaIR climate model, means that the method has been generalised to calculate warming contributions from any arbitrary combination of climate drivers, constrained for any time period, relative to any baseline, applying any definition of warming period, using any observation datasets, and for any alternative scenarios of historical and future forcings. Please get in touch with the authors if you are interested in applying the method in this way.

The following animation steps through the calculation of the Global Warming Index, showing how the different datasets and methods are combined to produce the final result. For simplicity, the 5-95% uncertainty plumes are shown for every step of the process.

Average difference between forced warming and observed warming
0.0000 K
1
2
3
4
5
6

Background to the Global Warming Index

The globalwarmingindex.org website shows an up-to-the-second index of human-induced warming relative to the second half of the 19th century (1850-1900).

The Global Warming Index was first introduced in Otto et al. (2015) based on a "detection and attribution" methodology originally developed by Hasselmann (1997). It was published with a comprehensive uncertainty analysis in Haustein et al. (2017), and work extending the robustness and utility of the Index is ongoing .

The Global Warming Index is been used extensively in the IPCC Assessment Reports to assess the level of global warming (including the Special Report on 1.5°C in 2018 , and the 6th Assessment Report in 2021 ), and for more broadly tracking progression towards the Paris Agreement goals.

Annual Assessments and Reach

The GWI is under active scientific development and maintenance by researchers from the Environmental Change Institute at the University of Oxford , UK, who provide systematic operarational annual updates of the datasets and methodology.

These latest GWI datasets are peer reviewed and published as part of the Indicators of Global Climate Change (IGCC) initiative global annual climate assessments, which are released for policymakers at the yearly UNFCCC Bonn meetings. These assessments have additionally been presented at the annual UNFCCC COP conferences, and included in the WMO State of the Climate Report.

The GWI is now widely used by policymakers, scientists, and climate communicators around the world to track progress towards international climate goals. As one example of reach, the IGCC's 2024 assessment was the most featured climate article in the media in 2025, according to Carbon Brief .

Usage of the GWI Data

To use the Global Warming Index data in publications, we kindly ask you to cite the following papers as appropriate:

Full List of Relevant Papers

Contributing authors

The contributing authors and their affiliations are as follows: