Incorporating feedback uncertainties in a model-based assessment of equilibrium climate sensitivity using Last Glacial Maximum temperature reconstructions

Abstract:

As the most recent period of large climate change, the Last Glacial Maximum (LGM) has been a useful target for analysis by model-data comparison. In addition, significant changes in greenhouse gas forcing across the last deglaciation and the relative wealth of LGM temperature reconstructions by proxy data provide a potentially useful opportunity to quantify equilibrium climate sensitivity (ECS), the change in global mean surface air temperature due to a doubling of atmospheric CO₂. Past model-data comparisons have attempted to estimate ECS using the LGM climate in two ways: (1) scaling of proxy data with results from general circulation model intercomparisons, and (2) comparing data with results from an ensemble of ECS-tuned simulations using a single intermediate complexity model. While the first approach includes the complexity of climate feedbacks, the sample size of the ECS-space may be insufficiently large to assess climate sensitivity. However, the second approach may be model dependent by not adequately incorporating uncertainty in climate feedbacks. Here, we present a new LGM-based assessment of ECS using the latter approach along with a simple linear parameterization of the longwave and shortwave cloud feedbacks derived from the CMIP5/PMIP3 results applied to the University of Victoria Earth System intermediate complexity model (UVIC). Cloud feedbacks are found to be the largest source of variability among the CMIP5/PMIP3 simulations, and our parameterization emulates these feedbacks in the UVIC model. In using this parameterization, we present a new ensemble of UVIC simulations to estimate ECS based on a Bayesian comparison with LGM temperature reconstructions that determines a probability distribution of optimal overlap between data and model results.