Parameter Uncertainty Quantification in an Idealized GCM With a Seasonal Cycle

Climate models are generally calibrated manually by comparing selected climate statistics, such as the global top-of-atmosphere energy balance, to observations. The manual tuning only targets a limited subset of observational data and parameters. Bayesian calibration can estimate model parameters their uncertainty using larger fraction available automatically exploring parameter space more broadly. In learning, it is natural exploit seasonal cycle, which has large amplitude compared with anthropogenic change in many statistics. this study, we develop methods for quantification (UQ) exploiting demonstrate proof-of-concept an idealized general circulation (GCM). UQ performed calibrate-emulate-sample approach, combines stochastic optimization machine learning emulation speed up learning. demonstrated perfect-model setting through convective parameterization GCM cycle. Calibration based on seasonally averaged annually averaged, reduces error order magnitude narrows spread non-Gaussian posterior distributions factors between two five, depending variables used UQ. reduction distribution leads predictions.