DE Simulations

The first funding phase of the HD(CP)² project was used to develop an ICON version that can be run in a high-resolution large eddy mode (ICON-LEM). The core ingredient for this was the ICON (ICOsahedral Non-hydrostatic atmosphere) model which has been developed in cooperation between MPI-M (Max-Planck Institute for Meteorology) and the DWD (Deutscher Wetterdienst). The project-own large eddy model is capable of performing hindcast simulations on domains as large as entire Germany (DE simulations) on a horizontal resolution of 652m, 312m, and 156m. Until now, model simulations of such detail have not been possible due to limitations in computing power.
In cooperation with the DKRZ (Deutsches Klimarechenzentrum) it is now possible to enable such model simulation not only on the DE domain, but also on other domains.

ICON-LEM Key Features

  • Hindcast simulation time: 24 hrs or longer
  • Core domain: DE (domain shift possible)
  • Scaling up to 458752 cores
  • 3 domain output: 625m, 312m, 156m horizontal resolution
  • Model output: 50 TB (74 variables) and 16 TB restart files

Model Evaluation

At the end of the first funding phase, the existing ICON-LEM version was tested against several observational data and against the established Opens external link in new windowCOSMO (COnsortium for Small-scale Modelling) model. Firstly, it was tested whether the model is able to correctly simulate the small- to mesoscale variability in turbulence, cloud formation and precipitation. Secondly, the results will be used to advance the understanding of moist processes in the atmosphere and their parametrization. The resulting publication, Heinze et al. (2017), is freely available Opens external link in new windowhere.

The detailed model evaluation showed that the ICON-LEM could well compete with the established COSMO model, even at an early development stage without extensive model tuning. Moreover, it showed that the high-resolution indeed yielded a significantly improved representation of small-scale variabilities.

The second funding phase of the HD(CP)² project will be used to apply this model to a variety of .

Back to top