Does Model Calibration Reduce Uncertainty in Climate Projections?

Uncertainty in climate projections is large as shown by the likely uncertainty ranges in Equilibrium Climate Sensitivity (ECS) of 2.5-4K and in the Transient Climate Response (TCR) of 1.4-2.2K. Uncertainty in model projections could arise from the way in which unresolved processes are represented, the parameter values used, or the targets for model calibration. We show that, in two climate model ensembles which were objectively calibrated to minimise differences from observed large scale atmospheric climatology, uncertainties in ECS and TCR are about two to six times smaller than in the CMIP5 or CMIP6 multi-model ensemble. We also find that projected uncertainties in surface temperature, precipitation and annual extremes are relatively small. Residual uncertainty largely arises from unconstrained sea-ice feedbacks. The 20+ year old HadAM3 standard model configuration simulates observed hemispheric scale observations and pre-industrial surface temperatures about as well as the median CMIP5 and CMIP6 ensembles while the optimised configurations simulates these better than almost all the CMIP5 and CMIP6 models. Hemispheric scale observations and pre-industrial temperatures are not systematically better simulated in CMIP6 than in CMIP5 though the CMIP6 ensemble seems to better simulate patterns of large-scale observations than the CMIP5 ensemble and the optimised HadAM3 configurations. Our results suggest that most CMIP models could be improved in their simulation of large scale observations by systematic calibration. However, the uncertainty in climate projections (for a given scenario) likely largely arises from the choice of parametrisation schemes for unresolved processes (“structural uncertainty”), with different tuning targets another possible contributor.

Quantifying non-CO2 contributions to remaining carbon budgets

The IPCC Special Report on 1.5 °C concluded that anthropogenic global warming is determined by cumulative anthropogenic CO2 emissions and the non-CO2 radiative forcing level in the decades prior to peak warming. We quantify this using CO2-forcing-equivalent (CO2-fe) emissions. We produce an observationally constrained estimate of the Transient Climate Response to cumulative carbon Emissions (TCRE), giving a 90% confidence interval of 0.26–0.78 °C/TtCO2, implying a remaining total CO2-fe budget from 2020 to 1.5 °C of 350–1040 GtCO2-fe, where non-CO2 forcing changes take up 50 to 300 GtCO2-fe. Using a central non-CO2 forcing estimate, the remaining CO2 budgets are 640, 545, 455 GtCO2 for a 33, 50 or 66% chance of limiting warming to 1.5 °C. We discuss the impact of GMST revisions and the contribution of non-CO2 mitigation to remaining budgets, determining that reporting budgets in CO2-fe for alternative definitions of GMST, displaying CO2 and non-CO2 contributions using a two-dimensional presentation, offers the most transparent approach.

On Spurious Causality, CO2, and Global Temperature

Stips et al. (2016) use information flows (Liang (2008, 2014)) to establish causality from various forcings to global temperature. We show that the formulas being used hinge on a simplifying assumption that is nearly always rejected by the data. We propose the well-known forecast error variance decomposition based on a Vector Autoregression as an adequate measure of information flow, and find that most results in Stips et al. (2016) cannot be corroborated. Then, we discuss which modeling choices (e.g., the choice of CO2 series and assumptions about simultaneous relationships) may help in extracting credible estimates of causal flows and the transient climate response simply by looking at the joint dynamics of two climatic time series.

Estimating Remaining Carbon Budgets Using Temperature Responses Informed by CMIP6

A remaining carbon budget (RCB) estimates how much CO2 we can emit and still reach a specific temperature target. The RCB concept is attractive since it easily communicates to the public and policymakers, but RCBs are also subject to uncertainties. The expected warming levels for a given carbon budget has a wide uncertainty range, which increases with less ambitious targets, i.e., with higher CO2 emissions and temperatures. Leading causes of RCB uncertainty are the future non-CO2 emissions, Earth system feedbacks, and the spread in the climate sensitivity among climate models. The latter is investigated in this paper, using a simple carbon cycle model and emulators of the temperature responses of the Earth System Models in the Coupled Model Intercomparison Project Phase 6 (CMIP6) ensemble. Driving 41 CMIP6 emulators with 127 different emission scenarios for the 21st century, we find almost perfect linear relationship between maximum global surface air temperature and cumulative carbon emissions, allowing unambiguous estimates of RCB for each CMIP6 model. The range of these estimates over the model ensemble is a measure of the uncertainty in the RCB arising from the range in climate sensitivity over this ensemble, and it is suggested that observational constraints imposed on the transient climate response in the model ensemble can reduce uncertainty in RCB estimates.

Estimating remaining carbon budgets using temperature responses informed by CMIP6

A remaining carbon budget (RCB) estimates how much CO2 we can emit and still reach a specific temperature target. The RCB concept is attractive since it easily communicates to the public and policymakers, but RCBs are also subject to uncertainties. The expected warming levels for a given carbon budget has a wide uncertainty range, which increases with less ambitious targets, i.e., with higher CO2 emissions and temperatures. Leading causes of RCB uncertainty are the future non-CO2 emissions, Earth system feedbacks, and the spread in the climate sensitivity among climate models. The latter is investigated in this paper, using a simple carbon cycle model and emulators of the temperature responses of the Earth System Models in the Coupled Model Intercomparison Project Phase 6 (CMIP6) ensemble. Driving 41 CMIP6 emulators with 127 different emission scenarios for the 21st century, we find almost perfect linear relationship between maximum global surface air temperature and cumulative carbon emissions, allowing unambiguous estimates of RCB for each CMIP6 model. The range of these estimates over the model ensemble is a measure of the uncertainty in the RCB arising from the range in climate sensitivity over this ensemble, and it is suggested that observational constraints imposed on the transient climate response in the model ensemble can reduce uncertainty in RCB estimates.

Bayesian estimation of Earth's climate sensitivity and transient climate response from observational warming and heat content datasets

Future climate change projections, impacts, and mitigation targets are directly affected by how sensitive Earth's global mean surface temperature is to anthropogenic forcing, expressed via the climate sensitivity (S) and transient climate response (TCR). However, the S and TCR are poorly constrained, in part because historic observations and future climate projections consider the climate system under different response timescales with potentially different climate feedback strengths. Here, we evaluate S and TCR by using historic observations of surface warming, available since the mid-19th century, and ocean heat uptake, available since the mid-20th century, to constrain a model with independent climate feedback components acting over multiple response timescales. Adopting a Bayesian approach, our prior uses a constrained distribution for the instantaneous Planck feedback combined with wide-ranging uniform distributions of the strengths of the fast feedbacks (acting over several days) and multi-decadal feedbacks. We extract posterior distributions by applying likelihood functions derived from different combinations of observational datasets. The resulting TCR distributions when using two preferred combinations of historic datasets both find a TCR of 1.5 (1.3 to 1.8 at 5–95 % range) ∘C. We find the posterior probability distribution for S for our preferred dataset combination evolves from S of 2.0 (1.6 to 2.5) ∘C on a 20-year response timescale to S of 2.3 (1.4 to 6.4) ∘C on a 140-year response timescale, due to the impact of multi-decadal feedbacks. Our results demonstrate how multi-decadal feedbacks allow a significantly higher upper bound on S than historic observations are otherwise consistent with.