In the context of rapid human-caused climate change, regular updates of the state of knowledge of current and future climate are needed. New statistical methods using observational constraints underpinned estimates of present-day human-induced warming and projected future warming in the most recent IPCC report. As time goes by, and new updated observational records become available, how should estimates of the current and projected human-caused climate change be updated? Here, we use a perfect model framework and show that incorporating observations from every new year in observationally constrained projections improves their accuracy, without causing major year-to-year spurious variability on outcomes. The forced warming estimated for the current year also exhibits high enough stability to be considered as a robust indicator of the state of the climate system.
As current warming is approaching the lower, 1.5 °C limit of the Paris Agreement long-term temperature goal, there is a growing appetite to understand how climate change is unfolding and how fast it actually changes, in near real time. From this perspective, regular updates to the current state of the climate at global, regional or national scales are extremely useful. Several organisations carry out regular climate monitoring on a global scale, describing temperature variations across the globe on a yearly, monthly or even daily basis. However, the observed temperature is the result of internal variability (up to a few tenths of a degree on the global average, and even more regionally, including interannual to multidecadal fluctuations) superimposed on top of the forced response. A specific estimation of the forced component is required to characterise climate change to date, and to compare its current state with long-term global warming levels used in international climate negotiations. Separation of human and natural forced drivers of climate change, as well as internal variability, is also useful for understanding recent observations of surface temperature change. It is also important to understand what are the implications of the forced response to date for future climate.
The 6th Assessment Report (AR6) of Working Group I (WGI) of the Intergovernmental Panel on Climate Change (IPCC) used the average warming over the last observed 10 or 20 years to characterise the current state of the climate system. In a context of rapid warming, this choice may not be optimal, as the estimated warming level lags behind the changes that have occurred to date. It also has policy and communication implications as the world is getting closer to the 1.5 °C global warming limit included in the Paris Agreement. A recent update estimated both the human-induced warming over the last 10 years ( + 1.19 [1.0 to 1.4] °C over 2014-2023, hereafter the 10-year estimate), and 1.31 [1.1 to 1.7] °C for the year 2023 (hereafter the 1-year estimate). The difference between these two estimates suggests a (human-induced) warming of about 0.1 °C higher in 2023 than in the preceding decade, 2014-2023, a result fully aligned with the estimated decadal global warming rate of about 0.26 °C/decade over 2014-2023. This difference raises the question: which of these two estimators is the most accurate at characterising the forced warming experienced at present?
Updating forced warming estimates is also applicable to projections of the future climate. Until AR6, climate projections for the 21st century and beyond typically relied on raw simulations from climate models, with no direct use of observations. The AR6 saw a change of approach, with a narrowing of the large range of CMIP6 climate responses based on historical observations and independently assessed climate sensitivity ranges grounded in multiple lines of evidence. Assessed global projections thus rely on observational constraints, i.e., the filtering of relevant climate responses on the basis of available observations. Through this mechanism, constrained climate projections also become a function of recent observations. This naturally raises the possibility of regular updates of observationally-constrained projections, alongside other key indicators based on observations. As a recent example, the very high global-mean surface temperature (GST) of the second half of 2023 and 2024 has raised fair questions about their long-term implications, and is already the subject of a large body of literature: are such warm years consistent with the previously estimated forced warming trajectory? or should this trajectory be revised upwards? These questions regarding the current and future climate response are of high policy relevance.
As the IPCC reports are usually published in cycles of 5-7 years, and the AR7 WGI Reports would be published only in the last years of this decade, more regular updates are needed. These could help authors to trace recent developments, to contextualise the latest observations, and to assess remaining carbon budgets with increased accuracy. A recent initiative provided annual updates of several key indicators of the state of the global climate system for the first time, covering both societal (e.g., greenhouse gas emissions) and physical aspects (e.g., updated estimates of global surface temperature, radiative forcings, total forced and human-induced warmings, Earth's energy imbalance). Updates for this set of indicators will be published annually and are presented in an accompanying dashboard to provide an up-to-date description of the state of the climate.
In order to support and extend this effort, we investigate two intertwined questions. First, is the estimated forced warming to date a robust indicator? Second, should long-term projections be constrained by the latest available observations? Here, we investigate the influence of year-to-year internal variability, both at the global and regional scales, to assess the strength and limitations of using these indicators in near real time.
Our analysis is based on an observational constraint method, called Kriging for Climate Change (KCC), that seamlessly estimates past, present and future forced warming in response to different emission scenarios (see Fig. 1 and "Methods"). This method involves mainly a Kalman filter (or Kriging) of a range of climate model projections. Its implementation requires a careful estimation of the uncertainties related to climate models and observations (see Methods). This technique was one of the techniques assessed and applied in the attribution and projection chapters of the AR6 and in subsequent updates. Nevertheless, key conclusions of this study are expected to also hold for other statistical methods used to constrain projections or estimate attributable warming.