Tradeoffs in Acceleration and Initialization of Superparameterized Global Atmospheric Models for MJO and Climate Science

Friday, 19 December 2014: 4:45 PM
Michael S Pritchard, University California Irvine, Carlsbad, CA, United States, Christopher Stephen Bretherton, University of Washington Seattle Campus, Seattle, WA, United States and Charlotte A DeMott, Colorado State University, Atmospheric Science, Fort Collins, CO, United States
New trade-offs are discussed in the cloud superparameterization approach to explicitly representing deep convection in global climate models. Intrinsic predictability tests show that the memory of cloud-resolving-scale organization is not critical for producing desired modes of organized convection such as the Madden-Julian Oscillation (MJO). This has implications for the feasibility of data assimilation and real-world initialization for superparameterized weather forecasting. Climate simulation sensitivity tests demonstrate that 400% acceleration of cloud superparameterization is possible by restricting the 32-128 km scale regime without deteriorating the realism of the simulated MJO but the number of cloud resolving model grid columns is discovered to constrain the efficiency of vertical mixing, with consequences for the simulated liquid cloud climatology. Tuning opportunities for next generation accelerated superparameterized climate models are discussed.