作者: Kuniko Yamazaki , Daniel J. Rowlands , Tolu Aina , Adam T. Blaker , Andy Bowery
DOI: 10.1002/JGRD.50304
关键词:
摘要: A number of studies have set out to obtain a range atmosphere and ocean model behavior by perturbing parameters in single climate (perturbed physics ensemble: PPE). Early used shallow layer slab or flux-adjusted coupled ocean-atmosphere models broad as characterized sensitivity. recent study reports relatively narrow sensitivities (2.2–3.2°C) PPE 35 without flux adjustment, raising the question whether previous ranges were an artifact use that not top-of-atmosphere (TOA) energy balance. Moreover, no experiment has reported large spread compared exhibited multi-model ensemble (MME) such Coupled Model Intercomparison Project phase 3 (CMIP3). In this work, we randomly perturb general circulation using space-filling design containing 10,000 combinations. The is run over distributed computing platform climateprediction.net under fixed pre-industrial forcing adjustment. We resample second, 20,000-member, with perturbations conditioned on TOA fluxes from first drift significantly realistic base state while targeting behavior. Models within targeted show regional control climates when CMIP3 ensemble, although there bias global mean surface temperature. predicted equilibrium substantially smaller than obtained but larger 35-model un-flux-adjusted mentioned above. Atlantic meridional overturning exhibits strength wide found ensemble. conclude adjustment pre-requisite for obtaining perturbed