Efficiently Constraining Climate Sensitivity with Ensembles of Paleoclimate Simulations

摘要: We use a recently-developed efficient probabilistic estimation technique to estimate the sensitivity of Earth’s temperature doubling atmospheric carbon dioxide. The method is based on ensemble Kalman filter which we apply CCSR/NIES/ FRCGC AGCM (the component MIROC 3.2) at T21L20 resolution coupled slab ocean. combines prior beliefs about model, with observational data, simultaneously 25 model parameters in an and objective manner. perform analysis investigate effect different assumptions regarding error, since this necessarily subjective input has not yet been well characterised. attempt validate resulting ensembles against out-of-sample data by comparing their hindcasts Last Glacial Maximum (LGM) paleoclimate proxy demonstrate through that our simulations are probably biased towards too high sensitivity. Within framework single-model experiment, show climate much greater than 6° Ci s hard reconcile record, 8°C seems virtually impossible. Our for most likely region 4.5°C. Although these results reasonably consistent widely accepted estimates sensitivity, they disagree some recent research suggested significant probability sensitivities excess values. These suggest paleoclimatic evidence could provide useful, albeit imprecise, constraint forecasts future change.