Space weather modeling and simulation
Some of the effects of space weather are annoying without being dangerous or directly damaging, e.g. delays imposed on aeromagnetic survey flights as a consequence of a highly disturbed geomagnetic field. Other space weather effects, however, can be dangerous and may cause direct damage like excessive electrical charging of spacecraft and intense geomagnetically induced currents in electric power networks and pipeline systems. Also, the quality of wireless communication and navigation may be affected, which can influence safety and security in certain cases.
Because of these effects, every day predictions about the effect of space weather on Earth are being published.
Currently, the task of a space weather forecaster is to synthesize incoming information (satellite in-situ data, satellite images, ground-based images) and to produce a handful of parameters (geomagnetic indices, flare probabilities, EUV flux) that characterize the space weather effects on Earth in the coming days. Simulation runs are used as a validation, but the accuracy of today’s predictions is low.
One of the limiting factors in the accuracy of space weather forecasting today is data scarcity. There will never be enough satellites and telescopes to cover the full heliosphere. In this project, we want to replace actual measurements by measurements made by virtual telescopes based on highly accurate astrophysical simulations. A virtual telescope observes the ongoing numerical simulation and produces, from a fixed point in space, a view on how the heliosphere looks like from that position. This would be the virtual equivalent of a set of spacecrafts like the STEREO mission.
To achieve these goals, we first need to produce highly accurate astrophysical simulations, and this is where the need for exascale computing arises. We first need to improve the astrophysical – in particular space plasma – models that are currently being employed for space weather simulations. In the simulation of plasma systems, models broadly fall into two separate categories: kinetic (microscopic) and fluid (macroscopic). To facilitate the simulation of full space weather events the microscopic and the macroscopic models must meet and must be coupled.
However, state-of-the-art kinetic models require two orders more resolution to reach the scale at which they become relevant to macro simulations. On the other hand, fluid descriptions, attempting to describe macroscopic behavior of plasmas, cannot be simulated at enough resolution to resolve those regions where small scale physics is important. It needs at least two orders of more resolution. The complexity of these multi-scale and multi-physics models requires exascale computing power for realistic simulations with a predictive value.
- MulticoreBSP software updated
- Talk on “Semi-automatic SIMD-efficient data layouts for object-oriented programs” at CHEP2013
- Seminar “Computational methods for modelling protein folding and aggregation” by Joost Schymkowitz (VIB SWITCH Laboratory) on 26 June, 2013
- Best paper award at ISPASS for Multi-Threaded Sampling paper
- PRACE Preparatory Access Grant Obtained