Super-parameterization in ocean modeling: Application to deep convection

We explore the efficacy of “super parameterization” (SP) in ocean modeling in which local 2-d non-hydrostatic plume-resolving fine-grained (FG) models are embedded at each vertical column of a coarse-grained (CG) hydrostatic model. A general multi-scale algorithm is described in which tendencies from the FG models are projected onto the CG model which in turn constrains the average state of the FG models, coupling the two models together. The approach is tested in the context of models of open ocean deep convection and compared with a pure hydrostatic, coarse resolution model using convective adjustment (HYD) and a full 3-d non-hydrostatic plume-resolving simulation (NH). The SP model is found to be greatly superior to HYD at much less computational cost than the fully non-hydrostatic calculation.

Research highlights
► A super-parameterized model (SP) of deep convection is evaluated.
► 2-d non-hydrostatic fine-grain models are embedded in a coarse hydrostatic model.
► Coupling between the 2 model types is formulated as a general multi-scale algorithm.
► We compare the SP model with hydrostatic (HYD) and non-hydrostatic (NH) 3-d models.
► The SP model is much more accurate than HYD and much less expensive than NH.