9.6 Variable-resolution modelling and extreme scaling applications with the Model for Prediction Across Scales (MPAS)

Heinzeller, Dominikus, Karlsruhe Institute of Technology (KIT), Germany, and University of Augsburg (UniA), Germany, Michael G. Duda, National Center for Atmospheric Research, Matthijs Kramer, Wageningen University (WUR), Netherlands, Hugo Hartmann, Wim van den Berg, MeteoGroup, Netherlands, Gert-Jan Steeneveld, WUR, and Harald Kunstmann, KIT and UniA

First, we evaluate MPAS for a number of key synoptic situations over Europe using several meshes with various refinement ratios and a uniform 3km mesh as reference. We compare this to the operational WRF setup of MeteoGroup Netherlands. The WRF model performs very similar to the observations in all cases, as does the global 3km MPAS run. The variable-resolution runs perform well in most of the test cases, but show deficiencies in a convective hail event over the Netherlands and for certain mesh configurations. We also discuss the computational requirements for the different MPAS meshes and compare it to the operational WRF setup at MeteoGroup.

Second, we present work towards the application of the atmospheric core of MPAS on current and future high performance computing systems for problems at extreme scale. We address the issue of massively parallel I/O by extending the model to support the highly scalable SIONlib library. Using global uniform meshes with a convection-permitting resolution of 2-3km, we demonstrate the ability of MPAS-A to scale out to half a million cores while maintaining a high parallel efficiency. We also demonstrate the potential benefit of a hybrid parallelisation of the code (MPI/OpenMP) on the latest generation of Intel's Many Integrated Core Architecture, the Intel Xeon Phi Knights Landing.