Parallel coupling framework and advanced time integration methods for detailed cloud processes in atmospheric models
Cloud processes represent one of the major uncertainties in current weather forecast, air quality and climate models. In the last years, a detailed treatment of clouds and aerosols in three-dimensional models has been applied for process studies only, but not for operational applications because of very high computational costs.
The objective of this project is to overcome this barrier and enable the practical application of such complex model systems. Therefore, a dynamic data structure framework, suitable parallelization techniques, and advanced numerical methods for the implementation of cloud processes in atmospheric model systems are developed. The different time scales, the heterogeneity in time and space, as well as the interactions of the considered processes are treated in an efficient manner. To adapt the time step size to the dynamics in different regions and the time scale of the considered processes, multirate integration methods are implemented. The framework FD4 (Four-Dimensional Distributed Dynamic Data Structures), developed in this project, provides a flexible coupling of cloud models to existing atmospheric models and is generally applicable to multiphase modeling. The use of FD4 as coupling framework will enable a more efficient parallelization and the decrease of the memory costs in comparison to naive approaches. Exemplarily, the developed libraries and time integration schemes are utilized to include a detailed spectral cloud microphysics model in the model system COSMO-MUSCAT (formerly LM-MUSCAT) and to enable its practical application.
The framework FD4 is freely available as open source software at the FD4 homepage.
Partner
Leibniz-Institut für Troposphärenforschung (IfT) Leipzig
ZIH-Kontakt
Duration
September 2007 - August 2010
Funding
DFG