Global warming results from anthropogenic greenhouse gas emissions which upset the delicate balance between the incoming sunlight, and the reflected and emitted radiation from Earth. The imbalance leads to energy accumulation in the atmosphere, oceans and land, and melting of the cryosphere, resulting in increasing temperatures, rising sea levels, and more extreme weather around the globe. Despite the fundamental role of the energy imbalance in regulating the climate system, as known to humanity for more than two centuries, our capacity to observe it is rapidly deteriorating as satellites are being decommissioned.

We present a reconstruction of the surface climate of the mid-Pliocene Warm Period (mPWP), specifically Marine Isotope Stage (MIS) KM5c or 3.205 Ma. We combine the ensemble of climate model simulations, which contributed to the Pliocene Model Intercomparison Project (PlioMIP), with compilations of proxy data analyses of sea surface temperature (SST). The different data sets we considered are all sparse with high uncertainty, and the best estimate of annual global mean surface air temperature (SAT) anomaly varies from 2.1 up to 4.8 °C depending on the data source. We argue that the latest PlioVAR analysis of alkenone data is likely more reliable than other data sets we consider, and using this data set yields an SAT anomaly of 3.9±1.1 °C, with a value of 2.8±0.9 °C for SST (all uncertainties are quoted at 1 standard deviation). However, depending on the application, it may be advisable to consider the broader range arising from the various data sets to account for structural uncertainty. The regional-scale information in the reconstruction may not be reliable as it is largely based on the patterns simulated by the models.

Observing climate feedbacks to long-term global warming, which are crucial climate regulators, is not feasible within the observational record. However, linking them to top-of-the-atmosphere flux variations in response to natural surface temperature fluctuations (internal variability feedbacks) is a viable approach. We explore the use of relating internal variability to forced climate feedbacks in models and applying the resulting relationship to observations to constrain forced climate feedbacks. Our findings reveal strong longwave and shortwave feedback relationships in models during the 14-year overlap with the Clouds and the Earth's Radiant Energy System (CERES) record. Yet, due to the weaker relationship between internal variability and forced climate longwave feedbacks, the net feedback relationship remains weak, even over longer periods beyond the CERES record. However, after about half a century, this relationship strengthens, primarily due to reinforcements of the internal variability and forced climate shortwave feedback relationship. We therefore explore merging the satellite records with reanalysis to establish an extended data record. The resulting constraint suggests a stronger negative forced climate net feedback than the model's distribution and an equilibrium climate sensitivity of about 2.59 K (1.95 to 3.12 K, 5 %–95 % confidence intervals). Nevertheless, this method does not account for certain factors like biogeophysical–chemical feedbacks, inactive on short timescales and not represented in most models, along with differences in historical warming patterns, which may lead to misrepresenting climate sensitivity. Additionally, continuous satellite observations until at least the mid-2030s are essential for using purely observed estimates of the net internal variability feedback to constrain the net forced climate feedback and, consequently, climate sensitivity.

Anthropogenic aerosol particles partially mask global warming driven by greenhouse gases, both directly by reflecting sunlight back to space and indirectly by increasing cloud reflectivity. In recent decades, emissions of anthropogenic aerosols have declined globally and at the same time shifted from the North American and European regions, foremost to Southeast Asia. Using simulations with the Max Planck Institute Earth System Model version 1.2 (MPI-ESM1.2), we find that the direct effect of aerosols has continued to increase despite declining emissions. Concurrently, the indirect effect has diminished in approximate proportion to the emissions. In this model, which employs parameterized aerosol effects with constant regional direct effect efficiency, the enhanced efficiency of aerosol radiative forcing in emissions is associated with less cloud masking, longer atmospheric residence times, and differences in aerosol optical properties.

Recent observation-based studies suggest that the Amazon rainforest has lost substantial resilience since 1990, indicating that the forest might undergo a critical transition in the near future due to global warming and deforestation. The idea is to use trends in a lag-1 auto-correlation of leaf density as an early-warning signal of an imminent critical threshold for rainforest dieback. Here we test whether the observed change in auto-correlations could arise from internal variability using historical and control simulations of nine sixth-generation Earth system model ensembles (Phase 6 of the Coupled Model Intercomparison Project, CMIP6). We quantify trends in the leaf area index auto-correlation from both models and satellite-observed vegetation optical depth from 1990 to 2017. Four models reproduce the observed trend with at least one historical realization whereby the observations lie at the upper limit of model variability. Three out of these four models exhibit similar behavior in control runs, suggesting that historical forcing is not necessary for simulating the observed trends. Furthermore, we do not observe a critical transition in any future runs under the strongest greenhouse gas emission scenario (SSP5-8.5) until 2100 in the four models that best reproduce the past observed trends. Hence, the currently observed trends could be caused simply by internal variability and, unless the data records are extended, have limited applicability as an early-warning signal. Our results suggest that the current rapid decline in the Amazon rainforest coverage is not foremost caused by global warming.

The paleoclimate record provides a test-bed in which climate models can be evaluated under conditions of substantial CO₂ change; however, these data are typically under-used in the process of model development and evaluation. Here, we use a set of metrics based on paleoclimate proxy observations to evaluate climate models under three past time periods. We find that the latest CMIP6/PMIP4 ensemble mean does a remarkably good job of simulating the global mean surface air temperatures of these past periods, and is improved on CMIP5/PMIP3, implying that the modern climate sensitivity of the CMIP6/PMIP4 model ensemble mean is consistent with the paleoclimate record. However, some models, in particular those with very high or very low climate sensitivity, simulate paleo temperatures that are outside the uncertainty range of the paleo proxy temperature data; in this regard, the paleo data can provide a more stringent constraint than data from the historical record. There is also consistency between models and data in terms of polar amplification, with amplification increasing with increasing global mean temperature across all three time periods. The work highlights the benefits of using the paleoclimate record in the model development and evaluation cycle, in particular for screening models with too-high or too-low climate sensitivity across a range of CO₂ concentrations.

The Earth's energy imbalance, i.e. the difference between incoming solar radiation and outgoing reflected and emitted radiation, is the one quantity that ultimately controls the evolution of our climate system. However, despite its importance, there is limited knowledge of the exact magnitude of the energy imbalance, and the small net difference of about 1 W m⁻² between two large fluxes (approximately 340 W m⁻²) makes it challenging to measure directly. There has recently been renewed interest in using wide-field-of-view radiometers on board satellites to measure the outgoing radiation, as a possible method for deducing the global annual mean energy imbalance. Here we investigate how to sample in order to correctly determine the global annual mean imbalance and interannual trends, using a limited number of satellites. We simulate satellites in polar (90° inclination), sun-synchronous (98°) and precessing orbits (73, 82°), as well as constellations of these types of satellite orbits. We find that no single satellite provides sufficient sampling, both globally and of the diurnal and annual cycles, to reliably determine the global annual mean. If sun-synchronous satellites are used, at least six satellites are required for an uncertainty below 1 W m⁻². One precessing satellite combined with one polar satellite results in root-mean-square errors of 0.08 to 0.10 W m⁻², and a combination of two or three polar satellites results in root-mean-square errors of 0.10 or 0.04 W m⁻², respectively. In conclusion, at least two satellites that complement each other are necessary to ensure global coverage and achieve a sampling uncertainty well below the current estimate of the energy imbalance.

Clouds affect the sensitivity of the climate system by changing their distribution, height, and optical properties under climate change. Although the precise magnitude remains uncertain, the direct cloud response to an external forcing is known to be destabilising. Additionally, clouds have a masking effect on CO₂ forcing and can influence other feedback mechanisms such as the surface albedo feedback. To understand the overall impact of clouds, we compute how much the equilibrium climate sensitivity (ECS) to a doubling of CO₂ changes when clouds are made transparent to radiation in an Earth system model (MPI-ESM1.2, the Max Planck Institute for Meteorology Earth System Model version 1.2). In practice, to stabilise the model climate at near-preindustrial temperatures, the solar constant was reduced by 8.8 %. Our experiments reveal that clouds exert a stabilising influence on the model, with a clear-sky ECS of 4.29 K, which is higher than the corresponding full-sky ECS of 2.84 K, contrasting with their direct destabilising effect. Detailed partial radiative perturbation diagnostics show that beyond directly amplifying warming by themselves, clouds also strengthen the negative lapse rate and positive water vapour feedbacks, while strongly damping the positive albedo feedback. These findings highlight the complex role of clouds in modulating climate sensitivity.

The best estimate of equilibrium climate sensitivity (ECS) constrained based on the instrumental record of historical warming becomes coherent with other lines of evidence when the dependence of radiative feedback on the pattern of surface temperature change (pattern effect) is incorporated. Pattern effect strength is usually estimated with atmosphere-only model simulations forced with observed historical sea-surface temperature (SST) and sea-ice change and constant pre-industrial forcing. However, recent studies indicate that pattern effect estimates depend on the choice of SST boundary condition dataset, due to differences in the measurement sources and the techniques used to merge and construct them. Here, we systematically explore this dataset dependency by applying seven different observed SST datasets to the MPI-ESM1.2-LR model covering 1871–2017. We find that the pattern effect ranges from -0.01±0.09 to 0.42±0.10 W m⁻² K⁻¹ (standard error), whereby the commonly used Atmospheric Model Intercomparison Project II (AMIPII) dataset produces by far the largest estimate. When accounting for the generally weaker pattern effect in MPI-ESM1.2-LR compared to other models, as well as dataset dependency and intermodel spread, we obtain a combined pattern effect estimate of 0.37 W m⁻² K⁻¹ [−0.14 to 0.88 W m⁻² K⁻¹] (5th–95th percentiles) and a resulting instrumental record ECS estimate of 3.2 K [1.8 to 11.0 K], which as a result of the weaker pattern effect is slightly lower and better constrained than in previous studies.

The use of paleoclimates to constrain the equilibrium climate sensitivity (ECS) has seen a growing interest. In particular, the Last Glacial Maximum (LGM) and the mid-Pliocene warm period have been used in emergent-constraint approaches using simulations from the Paleoclimate Modelling Intercomparison Project (PMIP). Despite lower uncertainties regarding geological proxy data for the LGM in comparison with the Pliocene, the robustness of the emergent constraint between LGM temperature and ECS is weaker at both global and regional scales. Here, we investigate the climate of the LGM in models through different PMIP generations and how various factors in the atmosphere, ocean, land surface and cryosphere contribute to the spread of the model ensemble. Certain factors have a large impact on an emergent constraint, such as state dependency in climate feedbacks or model dependency on ice sheet forcing. Other factors, such as models being out of energetic balance and sea surface temperature not responding below −1.8^∘C in polar regions, have a limited influence. We quantify some of the contributions and find that they mostly have extratropical origins. Contrary to what has previously been suggested, from a statistical point of view, the PMIP model generations do not differ substantially. Moreover, we show that the lack of high- or low-ECS models in the ensembles critically limits the strength and reliability of the emergent constraints. Single-model ensembles may be promising tools for the future of LGM emergent constraint, as they permit a large range of ECS and reduce the noise from inter-model structural issues. Finally, we provide recommendations for a paleo-based emergent constraint and notably which paleoclimate is ideal for such an approach.