Biases and parameterization formulation uncertainties in the representation of boundary layer clouds remain a leading source of possible systematic error in climate projections. Here we show the first results from simulations in a new experimental climate model, the “Ultra-Parameterized’’ Community Atmosphere Model, UPCAM. We have developed UltraParameterization (UP) through seed DOE SciDac funding as an unusually high-resolution implementation of cloud superparameterization (SP) in which thousands of embedded 2D cloud-resolving models are embedded in a host global climate model. In UP, the cloud-resolving arrays have been upgraded to include sufficient internal resolution to explicitly generate the turbulent eddies that form marine stratocumulus and trade cumulus clouds. This is computationally costly but complements other available approaches for studying low clouds and their climate interaction, by avoiding parameterization of the relevant scales. In Parishani et al. (2017) we demonstrate UP as a promising target for exascale computing and test its skill for low cloud simulation by comparing retrospective weather forecasts and multi-month climatological simulations against satellite data constraints. The results show that UP, while not without its own complexity trade-offs, produces encouraging improvements in the geographic and vertical structure of low clouds in present climate. This now paves the way for applying UP to study the low cloud response to surface warming in the near future.
Recent Comments