How and why to upgrade cloud microphysics in climate models

Modern climate models increasingly need to incorporate comprehensive cloud microphysics, because cloud-radiation feedbacks, which depend strongly on cloud microphysics, are critical for modeling both global climate sensitivity and regional climate change. Furthermore, precipitation processes and cloud-aerosol interactions, both of which are also inherently microphysical, are important aspects of climate variability. In this research, we first develop a set of new parameterizations incorporating recent research on cloud microphysics. We evaluate these parameterizations against observations using a single-column model, a major global climate model, and an operational global numerical weather prediction model. Our tests show that the microphysically detailed parameterizations produce more realistic simulations than traditional ones, in all three models. The improvements hold over a spectrum of time scales from daily to seasonal, and in a broad range of geographical regions and climate regimes. By using a combination of models for testing against observations, we can also diagnose the detailed causes of model errors. Finally, we discuss a promising new approach in which the cloud-radiation parameterization problem is treated stochastically. This approach replaces the prediction of fractional cloud cover with a prediction of the probability distribution functions of cloud properties, incorporating observed statistical characteristics of cloud fields.