| Summary: | Probabilistic climate assessments require robust characterizations of decision-relevant uncertainties. Reduced complexity climate models can be useful tools for quantifying uncertainty, given their flexibility, computational efficiency, and the ability to link these models with large-ensemble frameworks. Here, we develop and use the Hector-BRICK reduced complexity climate model in Bayesian calibrations that assimilate information from global observational data sets to estimate model parameters. These calibrations produce sets of model parameters that are consistent with observational constraints and account for correlations between those parameters. The calibrations produce posterior samples with effective sample sizes on the order of 103, allowing characterization of the tails of parametric uncertainty up to and beyond 90% credible intervals. We apply the model and its calibrations in novel ways to address highly relevant climate challenges. We start with a short introductory chapter discussing probabilistic climate projection, the Hector-BRICK model, and a method applying Hector-BRICK for probabilistic projections. Then we move to chapter two, addressing the impact of different observational constraints on key model parameters and model projections. We find that thermal expansion information influences equilibrium climate sensitivity estimates, while other constraints related to land ice have little impact. The thermal expansion constraint also affects the upper tails of sea level projections. While other studies have explored the complementary nature of temperature and thermal expansion information to understand the climate system, none have simultaneously accounted for centennial scale down to decadal scale temperature change simultaneous with ocean heat constraints, as is implicit in our calibration process. This work was published in Earth’s Future (Vega-Westhoff et al., 2019). Next, we analyze the effects of uncertainty in Earth’s climate sensitivity on sea level projections. We separate model results into high and moderate/low equilibrium climate sensitivity and explore the effects on projections, time horizons, and spatial patterns of sea-level change. Results show sea-level rise projections depend significantly on equilibrium climate sensitivity (and similarly on transient climate response), which can affect estimates of timing of threshold exceedances and regional assessments. The dependence is strongest in the upper tail of sea-level rise scenarios. This analysis can inform regional sea-level assessments tailored for high-impacts scenarios, such as estimating the likelihood of potential future threshold responses, as well as the magnitude and timing of onset on decadal timescales. It is also relevant to discussions of the evolution of CMIP models, which may be trending to higher climate sensitivities. This is among the first published analysis of the relationship between climate sensitivity and sea level projections, beyond sensitivity studies that do not ensure consistency with observational constraints. This work was published in Geophysical Research Letters in (Vega-Westhoff et al., 2020). Finally, we produce hybrid model-emulations of probabilistic sea level projections. These hybrid results take into account ensembles of climate model projections of temperature and ocean thermal expansion, while also sampling from Hector-BRICK calibrations. The upper tail of the warming for the latest model ensemble is shifted to higher values than the previous ensemble. These warmer temperatures lead to a wider upper bound for sea level, largely due to the Greenland contribution, whose median contribution increases with temperature while its upper tail also widens. This work helps to fill the gap between computationally expensive process-based models and simpler statistical estimation techniques based on Bayesian calibration with observational constraints. Results are well-suited for multi-sector analysis and systems that are particularly vulnerable to extreme and deeply uncertain sea-level rise scenarios. In short, we explore sensitivities of climate projections to constraints on calibration targets, key climate parameter values, and projection targets. Our exploration includes particular focus on upper tail, high-impact estimates.
|
|---|