Climate Modelling and Numerical Weather Prediction
Numerical Earth system models are indispensable tools to assess climate change. Comprehensive coupled models allow to study the manifold feedback mechanisms occurring in the atmosphere, biosphere, hydrosphere, and lithosphere.
In the SDL Climate Science we are mainly working with the atmospheric models ICON and WRF.
ICON
The ICON (ICOsahedral Nonhydrostatic) modeling framework is a joint project between the Deutscher Wetterdienst (DWD), the Max-Planck-Institute for Meteorology (MPI-M), the Deutsches Klimarechenzentrum (DKRZ) and the Karlsruhe Institute of Technology (KIT) for developing a unified next-generation global numerical weather prediction (NWP) and climate modeling system.
The work on ICON in the SDL Climate Science started with the HD(CP)² project.
Within this project the porting and optimization of the model took place to use efficiently the whole JUQUEEN machine (Heinze, 2016).
Hindcast simulations with high-resolutions over Germany were analyzed with respect to clouds and precipitation (Costa-Suros, 2020; Rybka, 2021).
Currently the SDL Climate Science is involved in the investigation of the modular supercomputing architecture (MSA) with the ICON model. The atmospheric part of ICON is ported to GPUs, but the ocean part is currently performing better on CPUs. Therefore we use the MSA approach for coupled simulations. The atmospheric part is running on the booster and the ocean and I/O part is running on the cluster of JUWELS.
Additionally the ICON model is used for scientific studies. We performed mountain wave (MW) resolving hindcasts over the Drake passage for a 10-day period in 2010 with numerous observed MW cases (Kruse, 2022).
WRF
Weather Research and Forecasting (WRF) is a sophisticated and open-source regional numerical weather predition model for research or operational prediction. WRF includes two dynamical cores, various physics configurations, data assimilation, coupling framework, and chemistry module with the support for parallel computation. WRF is currently operated in many national meteorological centers, including National Center for Environmental Prediction, private companies for public or commercial usage.
WRF is designed to perform with non-hydrostatic or hydrostatic simulations by finite difference method for atmosphere simulations, with realistic inputs or idealized condition. For different regional conditions, WRF can apply different radiation physics, microphysics, planetary boundary layer physics, land surface processes, or additional parameterizations for conducting operational results or studies in atmospheric science.
Performance analysis on the microphysics of WRF by the ensemble version of WRF (ESIAS-met) over the European domain (Domain Size: 180x180x49) from Lu et al. (2023)
A probabilistic simulation using the ESIAS-met by producing 32 ensembles along with 6 different physics options during the saharan dust event in Oct. 2018. The cloud mask from remote sensing observation is attached for comparison. The dark blue shows a high agreement to the prediction of clouds (work in progress).
A simulation of tropical cyclone Soudeler (2018) by WRF (V4.2). The quivers show the wind direction and wind speed (by length and color in the same time). The contours show the total pressure. (Work in progress)