Search

Emulators of Earth System Model (ESM) outputs have the potential to become a powerful tool for the impacts research community. If successful in emulating the needed variables at the required spatial and temporal resolution, they can supply impact model(er)s with the inputs necessary to explore how future water, energy, economic, and land systems evolve under a wide range of future scenarios. This complements the availability of ESM output currently limited to a small number of future scenarios - due to the computational costs of running fully coupled climate models - since emulators are by construction computationally efficient. STITCHES is an open-source climate emulator that can produce time series of multiple ESM variables, at the ESM original temporal and spatial resolution, by recombining existing model output into new scenarios on the basis of the scenario global temperature trajectory. In this talk, we will focus on how STITCHES can be used to explore new scenarios by filling in the space between existing scenarios. We will also discuss how insights from emulators such as STITCHES may be used to inform climate modeling centers on where to prioritize resources in future scenario simulations.

Carbon dioxide (CO2) concentrations have increased in the atmosphere as a direct result of human activity and are at their highest level over the last 2-3 million years, with profound impacts on the Earth system. However, the magnitude and future dynamics of land and ocean carbon sinks are not well understood; therefore, the amount of anthropogenic fossil fuel emissions that remain in the atmosphere (the airborne fraction) is poorly constrained. This work aims to quantify the sources and controls of atmospheric CO2, the fate of anthropogenic CO2 over time, and the trend and robustness of the airborne fraction. We use Hector v3.0, a coupled simple climate and carbon cycle model, with the novel ability to explicitly track carbon as it flows through the Earth system. We use a priori probability distribution functions for key model parameters in a Monte Carlo analysis of 10,000 coupled carbon-climate model runs from 1750 to 2300. Results are filtered for physical realism against historical observations and CMIP6 projection data, and we calculate the relative importance of parameters controlling how much CO2 ends up in the atmosphere. Unsurprisingly, we find that anthropogenic emissions are the dominant source of near- and long-term atmospheric CO2, composing roughly 45% of the atmosphere, which is consistent with observational studies of the airborne fraction. The overwhelming majority of model runs exhibited a negative trend in the airborne fraction from 1960-2020, implying that current-day land and ocean sinks are proportionally taking up more carbon than the atmosphere. Furthermore, when looking at the destination of anthropogenic fossil fuel emissions, only a quarter ends up in the atmosphere while more than half of emissions are taken up by the land sink on centennial timescales. This study evaluates the likelihood of airborne fraction trends and provides insights into the dynamics and destination over time of anthropogenic CO2 in the Earth system.

Climate model output emulation has long been attempted to support impact research, mainly to fill in gaps in the scenario space. Given the computational cost of running coupled earth system models (ESMs), which are usually the domain of supercomputers and require on the order of days to weeks to complete a century-long simulation, only a handful of different scenarios are usually chosen to externally force ESM simulations. An effective emulator, able to run on standard computers in times of the order of minutes rather than days could therefore be used to derive climate information under scenarios that were not run by ESMs. Lately, the necessity of accounting for internal variability has also made the availability of initial-condition ensembles, under a specific scenario, important, further increasing the computational demand. At least so far, emulators have been limited to simplified ESM-like output, either seasonal, annual, or decadal averages of basic quantities, like temperature and precipitation, often emulated independently of one another. With this work, we propose a more comprehensive solution to ESM output emulation. Our emulator, STITCHES, uses existing archives of earth system models' (ESMs) scenario experiments to construct ESM-like output under new scenarios or enrich existing initial-condition ensembles, which is what other emulators also aim to do. Importantly, however, STITCHES' output has the same characteristics of the ESM output it sets out to emulate: multivariate, spatially resolved, and high frequency, representing both the forced component and the internal variability around it. STITCHES extends the idea of time sampling – according to which climate outcomes are stratified by the global warming level at which they manifest themselves, irrespective of the scenario and time at which they occur – to the construction of a continuous history of ESM-like output over the whole 21st century, consistent with a 21st-century trajectory of global surface air temperature (GSAT) derived from the scenario that has been chosen as the target of the emulation. STITCHES does so by first splitting the target GSAT trajectory into decade-long windows, then matching each window in turn to a decade-long window within an existing model simulation from the available scenario runs according to its proximity to the target in absolute size of the temperature anomaly and its rate of change. A look-up table is therefore created of a sequence of existing experiment–time-window combinations that, when stitched together, create a GSAT trajectory “similar” to the target. Importantly, we can then stitch together much more than GSAT from these windows, i.e., any output that the ESM has saved for these existing experiment–time-window combinations, at any frequency and spatial scale available in its archive. We show that the stitching does not introduce artifacts in the great majority of cases (we look at temperature and precipitation at monthly frequency and on the native grid of the ESM and at an index of ENSO activity, the Southern Oscillation Index). This is true even if the criteria for the identification of the decades to be stitched together are chosen to work for a smoothed time series of annual GSAT, a result we expect given the larger amount of noise affecting most other variables at finer spatial scales and higher frequencies, which therefore are more “forgiving” of the stitching. We successfully test the method's performance over many ESMs and scenarios. Only a few exceptions surface, but these less-than-optimal outcomes are always associated with a scarcity of the archived simulations from which we can gather the decade-long windows that form the building blocks of the emulated time series. In the great majority of cases, STITCHES' performance is satisfactory according to metrics that reward consistency in trends, interannual and inter-ensemble variance, and autocorrelation structure of the time series stitched together. The method therefore can be used to create ESM-like output according to new scenarios, on the basis of a trajectory of GSAT produced according to that scenario, which could be easily obtained by a simple climate model. It can also be used to increase the size of existing initial-condition ensembles. There are aspects of our emulator that will immediately disqualify it for specific applications, like when climate information is needed whose characteristics result from accumulated quantities over windows of times longer than those used as pieces by STITCHES, droughts longer than a decade for example. But for many applications, we argue that a stitched product can satisfy the climate information needs of impact researchers. STITCHES cannot emulate ESM output from scenarios that result in GSAT trajectories outside of the envelope available in the archive, nor can it emulate trajectories with shapes different from existing ones (overshoots with negative derivative, for example). Therefore, the size and characteristics of the available archives of ESM output are the principal limitations for STITCHES' deployment. Thus, we argue for the possibility of designing scenario experiments within, for example, the next phase of the Coupled Model Intercomparison Project according to new principles, relieved of the need to produce a number of similar trajectories that vary only in radiative forcing strength but more strategically covering the space of temperature anomalies and rates of change.

Microbially-explicit models may improve understanding and projections of carbon dynamics in response to future climate change, but their fidelity in simulating global-scale soil heterotrophic respiration (RH ), a stringent test for soil biogeochemical models, has never been evaluated. We used statistical global RH products, as well as 7,821 daily site-scale RH measurements, to evaluate the spatio-temporal performance of one first-order decay model (CASA-CNP) and two microbially-explicit biogeochemical models (CORPSE and MIMICS) that were forced by two different input datasets. CORPSE and MIMICS did not provide any measurable performance improvement; instead, the models were highly sensitive to the input data used to drive them. Spatial variability in RH fluxes was generally well simulated except in the northern middle latitudes (~50°N) and arid regions; models captured the seasonal variability of RH well, but showed more divergence in tropic and arctic regions. Our results demonstrate that the next generation of biogeochemical models shows promise, but also needs to be improved for realistic spatio-temporal variability of RH . Finally, we emphasize the importance of net primary production, soil moisture, and soil temperature inputs, and that jointly evaluating soil models for their spatial (global scale) and temporal (site scale) performance provides crucial benchmarks for improving biogeochemical models.

The pre-disturbance vegetation characteristics that predict carbon (C) cycling responses to disturbance are not well known. To address this gap, we initiated the Forest Resilience Threshold Experiment, a manipulative study in which more than 3600 trees were stem girdled to achieve replicated factorial combinations of four levels (control, 45, 65, and 85% gross defoliation) of disturbance severity and two disturbance types (targeting upper or lower canopy strata). Applying a standardized stability framework in which initial C cycling resistance to disturbance was calculated as the first-year natural log response ratio of disturbance and control treatments, we investigated to what extent pre-disturbance levels of species diversity, aboveground woody biomass, leaf area index, and canopy rugosity—a measure of structural complexity—predict the initial responses of subcanopy light-saturated leaf CO2 assimilation (Asat), aboveground wood NPP (ANPPw), and soil respiration (Rs) to phloem-disrupting disturbance. In the year following stem girdling, we found that above-ground C cycling processes, Asat and ANPPw, were highly resistant to increases in disturbance severity, while Rs resistance declined as severity increased. Disturbance type had no effect on first-year resistance. Pre-disturbance aboveground woody biomass, and canopy rugosity were positive predictors of ANPPw resistance and, conversely, negatively related to Rs resistance. Subcanopy Asat resistance was not related to pre-disturbance vegetation characteristics. Stability of C uptake processes along with Rs declines suggest the net C sink was sustained in the initial months following disturbance. We conclude that biomass and complexity are significant, but not universal, predictors of initial C cycling resistance to disturbance. Moreover, our findings highlight the utility of standardized stability measures when comparing functional responses to disturbance.

Hector is a carbon/climate model capable of emulating Earth System Model outputs at the global scale and is able to reproduce historical observations well. Like other participating models of the Reduced Complexity Model Intercomparison Project, Hector is a computationally efficient source of climate projections and thus has a wide range of applications such as scenario generation, coupling with integrated assessment models, outreach, education, and policy making. Hector version 3 includes a number of new features: carbon tracking, permafrost, improved land-ocean warming contrast, and a web browser-accessible interface. Here we summarize these developments and discuss how they improve the model’s performance and broaden its potential user base.

The concept of stability is central to the study and sustainability of vital ecosystem goods and services as disturbances increase globally. While ecosystem ecologists, including carbon (C) cycling scientists, have long‐considered multiple dimensions of disturbance response, our discipline lacks an agreed‐upon analytical framework for characterizing multidimensional stability. Here, we advocate for the broader adoption of a standardized and normalized multidimensional stability framework for analyzing disturbance response. This framework includes four dimensions of stability: the degree of initial change in C fluxes (i.e., resistance); rate (i.e., resilience) and variability (i.e., temporal stability) of return to pre‐disturbance C fluxes; and the extent of return to pre‐disturbance C fluxes (i.e., recovery). Using this framework, we highlight findings not readily seen from analysis of absolute fluxes, including trade‐offs between initial and long‐term C flux responses to disturbance; different overall stability profiles among fluxes; and, using a pilot dataset, similar relative stability of net primary production following fire and insect disturbances. We conclude that ecosystem ecologists’ embrace of a unifying multidimensional stability framework as a complement to approaches focused on absolute C fluxes could advance global change research by aiding in the novel interpretation, comprehensive synthesis, and improved forecasting of ecosystems’ response to an increasing array of disturbances.

We consider the problem of estimating the ensemble sizes required to characterize the forced component and the internal variability of a range of extreme metrics. While we exploit existing large ensembles contributed to the CLIVAR Large Ensemble Project, our perspective is that of a modeling center wanting to estimate a-priori such sizes on the basis of an existing small ensemble (we use five members here). We therefore ask if such small-size ensemble is sufficient to estimate the population variance in a way accurate enough to apply a well established formula that quantifies the expected error as a function of n (the ensemble size). We find that indeed we can anticipate errors in the estimation of the forced component for temperature and precipitation extreme metrics as a function of n by applying the population variance derived by five members in the formula. For a range of spatial and temporal scales, forcing levels (we use RCP8.5 simulations), and both models considered here as our proof of concept, CESM1-CAM5 and CanESM2, it appears that an ensemble size of 20 or 25 members can provide estimates of the forced component for the extreme metrics considered that remain within small absolute and percentage errors. Additional members beyond 20 or 25 add only marginal precision to the estimate, which remains true when extreme value analysis is used. We then ask about the ensemble size required to estimate the ensemble variance (a measure of internal variability) along the length of the simulation, and – importantly – about the ensemble size required to detect significant changes in such variance along the simulation with increased external forcings. When an F-test is applied to the ratio of the variances in question, one estimated on the basis of only 5 or 10 ensemble members, one estimated using the full ensemble (up to 50 members in our study) we do not obtain significant results even when the analysis is conducted at the grid-point scale. While we recognize that there will always exist applications and metric definitions requiring larger statistical power and therefore ensemble sizes, our results suggest that for a wide range of analysis targets and scales an effective estimate of both forced component and internal variability can be achieved with sizes below 30 members. This invites consideration of the possibility of exploring additional sources of uncertainty, like physics parameter settings, when designing ensemble simulations.

Forested ecosystems represent a large yet uncertain fraction of the global terrestrial carbon sink. Their future state depends on a number of natural and anthropogenic influences; a particularly large uncertainty is how disturbance affects vegetation structure and ecosystem biogeochemistry. We used the Ecosystem Demography model to explore the ecological and biogeochemical consequences of disturbance as part of the Forest Resilience Threshold Experiment (FoRTE), a dual modeling and manipulative field experiment investigating the effects of disturbance at different severities on a century-old deciduous forest. The field component was conducted at the University of Michigan Biological Station (UMBS), where stem girdling was applied to achieve four different severity levels of disturbance (0, 45, 65, and 85% gross defoliation) before the 2019 growing season. Since then, we have tracked the subsequent changes in vegetation and biogeochemistry. The modeling component attempted to simulate the FoRTE disturbance treatments within its framework. While we were able to instantiate a forest in ED with a similar climatology, soil characteristics, disturbance history, and vegetation of UMBS, baseline ED is ultimately unable to reproduce the vegetation dynamics and carbon fluxes observed at the UMBS control plots. This is consistent with previous work where the model is not capable of matching observed carbon and vegetation dynamics. However, ED’s response to the disturbance treatments is consistent with observations from UMBS: in both the model and UMBS experimental results, we observed different resiliences and carbon cycle responses with respect to disturbance severity. These intriguing results point to both weaknesses and new possibilities in the modeling of ecosystems facing rising disturbances and climate change.

The terrestrial carbon cycle is a major source of uncertainty in climate projections. Its dominant fluxes, gross primary productivity (GPP), and respiration (in particular soil respiration, RS), are typically estimated from independent satellite-driven models and upscaled in situ measurements, respectively. We combine carbon-cycle flux estimates and partitioning coefficients to show that historical estimates of global GPP and RS are irreconcilable. When we estimate GPP based on RS measurements and some assumptions about RS:GPP ratios, we found the resulted global GPP values (bootstrap mean $${149}_{-23}^{+29}$$ 149 − 23 + 29 Pg C yr−1) are significantly higher than most GPP estimates reported in the literature ($${113}_{-18}^{+18}$$ 113 − 18 + 18 Pg C yr−1). Similarly, historical GPP estimates imply a soil respiration flux (RsGPP, bootstrap mean of $${68}_{-8}^{+10}$$ 68 − 8 + 10 Pg C yr−1) statistically inconsistent with most published RS values ($${87}_{-8}^{+9}$$ 87 − 8 + 9 Pg C yr−1), although recent, higher, GPP estimates are narrowing this gap. Furthermore, global RS:GPP ratios are inconsistent with spatial averages of this ratio calculated from individual sites as well as CMIP6 model results. This discrepancy has implications for our understanding of carbon turnover times and the terrestrial sensitivity to climate change. Future efforts should reconcile the discrepancies associated with calculations for GPP and Rs to improve estimates of the global carbon budget.

Simple climate models (SCMs) are computationally efficient and capable of emulating global mean output of more complex Earth system models (ESMs). In doing so, SCMs can play a critical role in climate research as stand‐ins for the computationally more expensive models, especially in studies involving low, spatial, and/or temporal resolution, providing more computationally efficient sources of climate data. Here we use Hector v2.5.0 to emulate the multiforcing historical and RCP scenario output for 31 concentration and seven emission‐driven ESMs. When calibrating Hector, sufficient calibration data must be used to constrain the model; otherwise, climate and/or carbon parameters affecting physical processes may be able to trade off with one another, allowing for solutions to use physically unreasonable fitted parameter values as well as limiting the application of the SCM as an emulator. We also present a novel methodology that uses the ESM range as a calibration data, which can be adopted when faced with missing variable output from a specific model.

The fortedata R package is an open data notebook from the Forest Resilience Threshold Experiment (FoRTE) – a modeling and manipulative field experiment that tests the effects of disturbance severity and disturbance type on carbon cycling dynamics in a temperate forest. Package data consist of measurements of carbon pools and fluxes and ancillary measurements to help analyze and interpret carbon cycling over time. Currently the package includes data and metadata from the first three FoRTE field seasons, serves as a central, updatable resource for the FoRTE project team, and is intended as a resource for external users over the course of the experiment and in perpetuity. Further, it supports all associated FoRTE publications, analyses, and modeling efforts. This increases efficiency, consistency, compatibility, and productivity while minimizing duplicated effort and error propagation that can arise as a function of a large, distributed and collaborative effort. More broadly, fortedata represents an innovative, collaborative way of approaching science that unites and expedites the delivery of complementary datasets to the broader scientific community, increasing transparency and reproducibility of taxpayer-funded science. The fortedata package is available via GitHub: https://github.com/FoRTExperiment/fortedata (last access: 19 February 2021), and detailed documentation on the access, used, and applications of fortedata are available at https://fortexperiment.github.io/fortedata/ (last access: 19 February 2021). The first public release, version 1.0.1 is also archived at https://doi.org/10.5281/zenodo.4399601 (Atkins et al., 2020b). All data products are also available outside of the package as .csv files: https://doi.org/10.6084/m9.figshare.13499148.v1 (Atkins et al., 2020c).

The relatively short atmospheric lifetimes of methane (CH4) and black carbon (BC) have focused attention on the potential for reducing anthropogenic climate change by reducing Short-Lived Climate Forcer (SLCF) emissions. This paper examines radiative forcing and global mean temperature results from the Energy Modeling Forum (EMF)-30 multi-model suite of scenarios addressing CH4 and BC mitigation, the two major short-lived climate forcers. Central estimates of temperature reductions in 2040 from an idealized scenario focused on reductions in methane and black carbon emissions ranged from 0.18–0.26 °C across the nine participating models. Reductions in methane emissions drive 60% or more of these temperature reductions by 2040, although the methane impact also depends on auxiliary reductions that depend on the economic structure of the model. Climate model parameter uncertainty has a large impact on results, with SLCF reductions resulting in as much as 0.3–0.7 °C by 2040. We find that the substantial overlap between a SLCF-focused policy and a stringent and comprehensive climate policy that reduces greenhouse gas emissions means that additional SLCF emission reductions result in, at most, a small additional benefit of ~ 0.1 °C in the 2030–2040 time frame.

Surface and space-based observations, field experiments, and models all contribute to our evolving understanding of the ways that Earth’s many systems absorb and release carbon.

Here we present results from the first phase of the Reduced Complexity Model Intercomparison Project (RCMIP). RCMIP is a systematic examination of reduced complexity climate models (RCMs), which are used to complement and extend the insights from more complex Earth System Models (ESMs), in particular those participating in the Sixth Coupled Model Intercomparison Project (CMIP6). In Phase 1 of RCMIP, with 14 participating models namely ACC2, AR5IR (2 and 3 box versions), CICERO-SCM, ESCIMO, FaIR, GIR, GREB, Hector, Held et al. two layer model, MAGICC, MCE, OSCAR and WASP, we highlight the structural differences across various RCMs and show that RCMs are capable of reproducing global-mean surface air temperature (GSAT) changes of ESMs and historical observations. We find that some RCMs are capable of emulating the GSAT response of CMIP6 models to within a root-mean square error of 0.2 °C (of the same order of magnitude as ESM internal variability) over a range of scenarios. Running the same model configurations for both RCP and SSP scenarios, we see that the SSPs exhibit higher effective radiative forcing throughout the second half of the 21st Century. Comparing our results to the difference between CMIP5 and CMIP6 output, we find that the change in scenario explains approximately 46 % of the increase in higher end projected warming between CMIP5 and CMIP6. This suggests that changes in ESMs from CMIP5 to CMIP6 explain the rest of the increase, hence the higher climate sensitivities of available CMIP6 models may not be having as large an impact on GSAT projections as first anticipated. A second phase of RCMIP will complement RCMIP Phase 1 by exploring probabilistic results and emulation in more depth to provide results available for the IPCC's Sixth Assessment Report author teams.

Typically in human-earth system modeling studies, feedbacks between the earth and human systems are analyzed by passing information between independent models. The reliance on existing Earth System Model outputs limits the ability to explore feedbacks under arbitrary scenarios and equally important limits the ability to explore large-scale uncertainty in these interactions. In this study we explore a wide range of climate uncertainties and incorporate the implications of increased cooling hydrofluorocarbons emissions. We implement a statistical relationship between global mean temperature change and heating and cooling degree days that allows us to produce changes in building energy demands within GCAM at every time step and every region. While there is a general agreement in the literature that increasing temperatures will increase cooling energy demands and decrease heating energy demands, there has been no fully-coupled analysis of this dynamic that would, for example, account for the feedbacks on hydrofluorocarbons from increased cooling demands. The variation in the spatial distribution of temperatures leads to substantial variation in the change in cooling and heating energy across regions, with regions like USA, India and Sub-Saharan Africa experiencing a factor of two difference in cooling demands. While the feedbacks between building energy demand and global mean temperature are modest by themselves, this study prompts future research on coupled human-earth system feedbacks, in particular in regards to land, water, and other energy infrastructure.

Earth System Models (ESMs) are excellent tools for quantifying many aspects of future climate dynamics but are too computationally expensive to produce large collections of scenarios for downstream users of ESM data. In particular, many researchers focused on the impacts of climate change require large collections of ESM runs to rigorously study the impacts to both human and natural systems of low-frequency high-importance events, such as multi-year droughts. Climate model emulators provide an effective mechanism for filling this gap, reproducing many aspects of ESMs rapidly but with lower precision. The fldgen v1.0 R package quickly generates thousands of realizations of gridded temperature fields by randomizing the residuals of pattern scaling temperature output from any single ESM, retaining the spatial and temporal variance and covariance structures of the input data at a low computational cost. The fldgen v2.0 R package described here extends this capability to produce joint realizations of multiple variables, with a focus on temperature and precipitation in an open source software package available for community use (https://github.com/jgcri/fldgen). This substantially improves the fldgen package by removing the requirement that the ESM variables be normally distributed, and will enable researchers to quickly generate covarying temperature and precipitation data that are synthetic but faithful to the characteristics of the original ESM.

Energy-economic models are used to provide science-based decision support in many contexts; evaluating scenarios generated by these models is critical for establishing the credibility of the analyses these tools support. We propose a framework for evaluating energy system scenarios consisting of three parts – a qualitative storyline, quantitative metrics, and evaluation criteria. We apply this framework to the reference scenario for GCAM-USA, a version of the Global Change Analysis Model with state-level detail in the United States, focusing on the evolution of the electric power sector. We develop new visual analytic tools to facilitate evaluation of model outcomes in 51 sub-national regions, and demonstrate how scenario performance can be tracked and compared across four quantifications of the GCAM-USA reference scenario.

The increasing data requirements of complex models demand robust, reproducible, and transparent systems to track and prepare models’ inputs. Here we describe version 1.0 of the gcamdata R package that processes raw inputs to produce the hundreds of XML files needed by the GCAM integrated human-earth systems model. It features extensive functional and unit testing, data tracing and visualization, and enforces metadata, documentation, and flexibility in its component data-processing subunits. Although this package is specific to GCAM, many of its structural pieces and approaches should be broadly applicable to, and reusable by, other complex model/data systems aiming to improve transparency, reproducibility, and flexibility. Funding statement: Primary support for this work was provided by the U.S. Department of Energy, Office of Science, as part of research in Multi-Sector Dynamics, Earth and Environmental System Modeling Program. Additional support was provided by the U.S. Department of Energy Offices of Fossil Energy, Nuclear Energy, and Energy Efficiency and Renewable Energy and the U.S. Environmental Protection Agency.

Over the last decades, climate science has evolved rapidly across multiple expert domains. Our best tools to capture state‐of‐the‐art knowledge in an internally self‐consistent modeling framework are the increasingly complex fully coupled Earth System Models (ESMs). However, computational limitations and the structural rigidity of ESMs mean that the full range of uncertainties across multiple domains are difficult to capture with ESMs alone. The tools of choice are instead more computationally efficient reduced complexity models (RCMs), which are structurally flexible and can span the response dynamics across a range of domain‐specific models and ESM experiments. Here we present Phase 2 of the Reduced Complexity Model Intercomparison Project (RCMIP Phase 2), the first comprehensive intercomparison of RCMs that are probabilistically calibrated with key benchmark ranges from specialized research communities. Unsurprisingly, but crucially, we find that models which have been constrained to reflect the key benchmarks better reflect the key benchmarks. Under the low‐emissions SSP1‐1.9 scenario, across the RCMs, median peak warming projections range from 1.3 to 1.7°C (relative to 1850–1900, using an observationally based historical warming estimate of 0.8°C between 1850–1900 and 1995–2014). Further developing methodologies to constrain these projection uncertainties seems paramount given the international community's goal to contain warming to below 1.5°C above preindustrial in the long‐term. Our findings suggest that users of RCMs should carefully evaluate their RCM, specifically its skill against key benchmarks and consider the need to include projections benchmarks either from ESM results or other assessments to reduce divergence in future projections., Key Points Probabilistic global‐mean temperature projections often use reduced complexity climate models (RCMs) because of their low computational costWe evaluate how well RCMs’ probabilistic setups can synthesize knowledge from multiple research domains for policy relevant projectionsNo RCM is able to capture all forcing, warming, heat uptake and carbon cycle metrics we evaluate, however some come close across a range