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

Abstract

Abstract. 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 effective climate sensitivity (ECS) and transient climate response (TCR). However, the ECS 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 ECS and TCR by using historic observations of surface warming, since the mid-19th century, and ocean heat uptake, 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 slow feedbacks (acting over decades). We extract posterior distributions by applying likelihood functions derived from different combinations of observational datasets. The resulting TCR distributions are similar when using different historic datasets: from a TCR of 1.5 (1.3 to 1.7 at 5–95 % range) °C, up to 1.7 (1.4 to 2.0) °C. However, the posterior probability distribution for ECS on a 100-year response timescale varies depending on which combinations of temperature and heat content anomaly datasets are used: from ECS of 2.2 (1.5 to 4.5) °C, for datasets with less historic warming, up to 2.8 (1.8 to 6.1) °C, for datasets with more historic warming. Our results demonstrate how differences between historic climate reconstructions imply significant differences in expected future global warming.

Climate nonlinearities: selection, uncertainty, projections, & damages

“…prior). This is in agreement with the finding in [10] that the most negative prior aerosol forcing values were excluded from the posterior (Fig. 3 therein).…”

“…While we have focused on arguably the simplest representation of climate physics, even this single-equation energy-balance model is used in a variety of applications for which structural uncertainty has real physical and economic consequences. Our approach to characterising structural uncertainty is equally applicable to more complex representations of climate physics, such as the model used in [10]; if such a model is computationally efficient enough to generate an ensemble that captures parametric uncertainty, one can also use model selection parameters for its different components (e.g. a parameter choosing between several representations of the ocean circulation) in the same fashion.…”

“…Thus the Intergovernmental Panel on Climate Change largely relies upon a heuristic characterisation of uncertainties [17]. On the other hand, simpler models of climate physics, such as the energy balance models (EBMs) used within the integrated assessment models (IAMs) of economists [4], or efficient reduced-complexity ESMs [10], can produce large ensembles of simulations, intended to characterise this uncertainty. While ESMs are typically the core tool for of climate science projections such as in [17], EBMs are used to emulate ESMs, to explore alter-native scenarios, assess parametric uncertainty, constrain with observations, and to represent climate physics within climate economics; thus both model frameworks are fundamentally important in guiding climate policy.…”

“…τ dt ′ (slow feedback) where the dot represents a time derivative, c [J/m 2 K] is the heat capacity of the Earth's surface layer represented by T , λ [W/m 2 K] is the climate feedback parameter, λ F,T and c F,T are nonlinear terms corresponding to quadratic Taylor expansions of the linear model [3], and λ s is a slow feedback term representing a multidecadal feedback on timescale τ [10]. By introducing a model selection parameter µ we can describe these in a single multimodel equation:…”

Probabilistic projections of baseline twenty-first century CO2 emissions using a simple calibrated integrated assessment model

“…As a result, these projections are best understood as a reflection of what CO 2 emissions might look like without significant changes in mitigation policy implementation or technological development and deployment patterns. We refrain from projecting global mean temperatures, as we do not consider the effects of carbon-cycle and biogeochemical dynamics and their uncertainties, which can have a large impact on the resulting CO 2 concentrations (Booth et al 2017 ; Quilcaille et al 2018 ), as well as uncertainties related to climate sensitivity (Goodwin and Cael 2021 ).…”