Confronting Climate Model Simulations with Satellite-Based Evaluation of Warm Rain Formation: Can We Reconcile "Bottom-up" Process-Based Constraints with the "Top-Down" Temperature Trend Constraints?

Cloud parameterizations in climate models include a number of adjustable parameters that arise from uncertainties in cloud processes. These parameters are often tuned to best reproduce specific aspects of the observed climate, such as the energy balance at the top of the atmosphere. Starting with the CMIP5 GFDL CM3 coupled climate model, we construct alternate model configurations that achieve the desired energy balance using different, but plausible, combinations of parameters. The present-day climate is nearly indistinguishable in all configurations, but the evolution of the surface temperature from pre-industrial to present-day differs markedly among these configurations due to a large spread in the magnitude of the aerosol indirect effect. Details of the cloud-to-rain conversion processes are found to be the source for this large spread.

Recently developed methodologies to analyze the CloudSat and A-Train satellite observations are employed to construct the statistical "fingerprint" process-level signatures of the cloud-to-rain processes. These methodologies are applied to both satellite observations and climate models. Such comparisons can be used to help constrain uncertain parameters included in cloud parameterizations. One of the highlighted results demonstrates that the model predictability of twentieth-century historical temperature trends contradicts the process-based constraint on a tunable cloud parameter. This implies the presence of compensating errors at a fundamental level, and underscores the importance of observation-based, process-level constraints on model microphysics uncertainties for more reliable predictions of the aerosol indirect effect. This uncertainty in the magnitude of the aerosol indirect effect ultimately limits our ability to constrain the climate sensitivity.