Organized convective systems in the tropical western pacific as a process in general circulation models: A toga coare case-study



We examine the large-scale effects of organized convective systems in the tropical western Pacific observed during the Tropical-Ocean Global-Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA COARE). In a case-study approach, we examine realizations of a supercluster, associated with the onset of the December 1992 westerly wind burst, in the T213 operational medium-range weather forecasting model of the European Centre for Medium-Range Weather Forecasts (ECMWF). We idealize a supercluster as a hierarchy of three interacting scales, namely organized cumulonimbu ��1 mesoscale convective systems ��2, and the supercluster component ��3. It is shown that the ECMWF model represents this hierarchy as a ��3-like surrogate whose influence dominates the effect of parametrized convection. This causes over-prediction of the model tendencies which, in the case of zonal momentum, is explained in elementary terms.

The structure of the resolved-scale momentum flux is explained by Moncrieff's (1992) archetypal theory of organized convection which has been verified against observations and cloud-resolving model data-sets. the parametrization of subgrid-scale convective momentum-flux in the ECMWF model, based on a momentum mixing concept, produces subgrid-scale tendencies that are physically different from transports associated with cumulonimbus convection in a shear flow.

We outline a strategy for parametrizing the momentum flux by the ��1 component based on the archetypal model. the ��2 component, which is part-resolved and part-parametrized, is at odds with the assumptions of scale separation underpinning parametrization. It is argued that this component should be represented as part of the prognostic treatment of convectively generated cirrus.

Finally, we suggest cloud-resolving modelling studies to further quantify the structure and large-scale impact of superclusters in a westerly-wind-burst environment, ranging from idealized models to models having data assimilation capability.