Modelling the Near-Earth Space in Six Dimensions
Finnish Centre of Excellence in Research of Sustainable Space, Helsinki (Finland)
Local Project ID:
HPC Platform used:
Hawk at HLRS
Space is the finest plasma laboratory one can reach, and hence many of the fundamental and universal physics discoveries of to the fourth state of matter – plasma – root to space physics. The near-Earth space is the only place one can send spacecraft to study the variability of plasma that ranges from meters to millions of kilometres and from milliseconds to hundreds of years. There is one but: Normally one can send only a few satellites, leaving gaps in observations. That is why near-Earth space environment modelling is so crucial.
Modelling space plasmas has three broad categories from computationally efficient to almost impossible. The most widely used option is to assume that plasmas are not electrons and protons, but form a fluid. In this case, the modeler chooses a volume to be simulated, and fills it with relatively coarse grid, where each cell is like pixels in a 3D camera picture. The computationally most difficult is to model electrons and protons as particles, in which case the simulation volume needs to be filled with tiny grid cells because electron-scale physics occurs in very small spatial and temporal scales. Since space is big – easily the volume to be simulated is million times million times million kilometres – electron-scale physics can only be carried out in small volumes now and in the foreseeable future.
However, there is a midway, in which protons are particles and electrons are fluid. Even this hybrid method is so demanding computationally that it has been feasible only in two spatial dimensions (see e.g., https://www.gauss-centre.eu/results/astrophysics/article/global-kinetic-modelling-of-space-weather-with-extreme-scalability-vlasiator/). Until now: A Finnish group was able to extend the world’s most accurate space environment simulation Vlasiator to cover six dimensions.
Umm – six dimensions? If you measure the temperature of plasmas in space, you often do not get a nice normal distribution of particles accumulating around one temperature as in non-magnetized gas, like air. You might get three or more different temperatures per location, meaning that to model the plasma temperature – and by extension almost everything that matters in plasmas – you need to model how the particles are distributed. This needs an additional three-dimensional space, inside the three-dimensional position space that contains the model grid pixels. The fluid models are computationally feasible exactly because they omit this step and assume that plasmas are like air and have only one temperature. This assumption makes the fluid models to hold only in the big picture, while many of the dynamical features made by the particles are lost.
To model ion-scale physics within the vast near-Earth space therefore requires a good resolution for the 3D position space, and additional 3D space for particle distributions. If the simulations are carried out for a long physical time period, the computational demands easily reach the performance capabilities of any available supercomputer – hence Vlasiator production runs are always ran with the finest possible machines. HLRS machines ranging already from the legendary Hermit to Hornet, Hazel Hen and now the Hawk belong to the group’s absolute favourites: reliable computing with knowledgeable first-aid always handy.
The latest runs on Hawk were focussing on one of the most mysterious questions in space physics: what causes the Earth’s magnetospheric tail to erupt plasma clouds at times, similar to Solar flares? This is a question that spacecraft haven’t been able to answer, because the process occurs in a large volume within a short period of time. It’s also a question which hasn’t been answered by previous 3D models, because the physics occurs in ion-scales, not in the fluid scales that have been targeted by the earlier modelling efforts. The results are absolutely fascinating, and the group is now researching in detail what makes the Earth send its own eruptions – stay tuned for the coming results!
Professor in Computational Space Physics
Director, Finnish Centre of Excellence in Research of Sustainable Space
University of Helsinki, Department of Physics, Room Exactum B409
also at: Finnish Meteorological Institute
P.O.Box 68, FI-00014 University of Helsinki (Finland)
e-mail: minna.palmroth [@] helsinki.fi
NOTE: This simulation project was made possible by PRACE (Partnership for Advanced Computing in Europe) allocating a computing time grant on GCS HPC system Hawk of the High-Performance Computing Center Stuttgart (HLRS). GCS is a hosting member of PRACE.
Local project ID: SIMPLE