Development of a new hybrid cumulus parameterization scheme for use in nonhydrostatic weather prediction models
Zusammenfassung der Projektergebnisse
Cumulus convection is one of the most energetic weather phenomena and plays an important role in the atmospheric heat, moisture and momentum transport. Since it is often anked by severe weather (heavy rain, hail, gusts, lightning), there is a strong interest in a reliable forecast by numerical weather prediction (NWP) models. Whereas in former NWP models with grid sizes delta x of several ten kilometers the spatial scales of convection were well in the subgrid range, this has changed over the years with ever decreasing delta x. Due to its subgrid scale character convection has to be parameterized, but for contemporal NWP models with delta x of only a few kilometers this has become problematic, because convection now is partially resolved. Especially the classical parameterization approach of a convective mass circulation closed within a single grid column has become questionable. To overcome this conceptual problem we have developed a hybrid mass flux convection scheme (HYMACS) which parameterizes only the small scale updrafts and downdrafts in a convective cell as subgrid processes and hands over the larger scale environmental subsidence to the grid scale dynamics of the hosting NWP model. The mass exchange between HYMACS and the hosting model leads to pressure gradient forces driving the environmental subsidence and producing more realistic dynamics in the environment of a convective cell. This dynamical HYMACS core then has been extended to a full cumulus convection scheme. A cloud model including water vapor, cloud water and ice, rain and snow with details adopted from existing convection schemes has been added. A cloud depth criterion allows for the representation of both deep and shallow convection by the same scheme. For consistency reasons with the hybrid dynamical approach a horizontal mass flux convergence closure has been chosen. The classical Fritsch-Chappell trigger evaluating grid scale vertical wind has been extended by a turbulent kinetic energy contribution and a non-local gust front contribution due to possibly existing upstream convective cells. These subgrid scale forcings led to more realistic precipitation patterns. The introduction of an individual age for each convective cell to parameterize a life cycle with a gradual cloud growth and precipitation evolution improved the cloud patterns, e.g. frontal precipitating clouds with leading younger shallow convection. Finally, the convective fractional cloud cover has been diagnosed to be coupled to the radiation scheme of the hosting model and convective momentum transport has been implemented. HYMACS has comprehensively been tested in the COSMO model of the Deutscher Wetterdienst, although its formulation also allows for the implementation in other non-hydrostatic NWP models. Idealized simulations ranging from single cells to sea-breaze convection and real case simulations covering both frontal and air mass convection proved HYMACS to be able to realistically represent convective dynamics, precipitation and cloud patterns for delta x of several ten down to about one kilometer, often superior to the classical convection schemes.
