In phase I, the HD(CP)² model was developed from an existing ICON non-hydrostatic core and model optimization started. Now, in the second phase of the project the focus will be on adapting the model to different next generation high-performance computing hardware. This is done in order to improve the computational efficiency (i.e. time to solution) and is fostered by the collaboration between the DKRZ (Deutsches Klimarechenzentrum) and the FZJ (Forschungszentrum Jülich). This topic will be addressed in work package 1 "Optimization of sustained application performance".
A central task of work package 2 is the implementation of a new subgrid cloud cover scheme to enable the experiments "Tropical Atlantic (TA)" and "North Atlantic (NA)". Moreover, the model will be enhanced by implementing and improving online diagnostic functionality (work package 3). We will conduct production simulations with this large-eddy model (LEM), therefore the current HD(CP)² ICON model is referred to as ICON_LEM. Simulations will be done as requested by the six scientific projects of HD(CP)² with resolution of O(100 m) over entire Germany ("D"), as well as - in the later stage of Phase II - NA and TA (work package 5).
Work package 4 will continue developing the end-to-end workflow of HD(CP)² and adapting it to support an automated execution chain, pre- and post-processing tools and basic data management. The scientific projects will be assisted in using these tools to conduct additional model runs.The resource planning is done in work package 6.
The HD(CP)² model and its scheduled simulations will be extremely ambitious regarding computing resources and data space. Moreover, model enhancement will include the implementation of advanced online diagnostic tools which enable a 3D visualization of features directly during the experiment runs.This is addressed in work package 3 "Model diagnostics".
Due to its remarkable scaling characteristics that were established in the first funding period of HD(CP)², model has the potential to be further developed into a demonstrator for extreme scale computing.
Work package 2
Work package 6