J1.4 Space weather forecasting with petascale computing (2008

J1.4

Space weather forecasting with petascale computing

Michael Wiltberger, NCAR, Boulder, CO

Moore's law is the well known empirical relationship stating that the number

of transistors on a integrated circuit doubles every 2 years leading to

dramatic increase computational power. During the GOES-R era this trend

implies that we will have routine access to petaFLOP, 10^15 FLoating point

OPerations per second, supercomputers. These computers will contain tens of

thousands of processors tied together with a sophisticated high speed network

which provide new challenges and opportunities for space weather forecasting.

One key challenge that must be addressed is writing numerical simulations

which can run on that many processors efficiently in order to reach the

unprecedented levels of resolution. One intriguing possibility is the

simulation of active regions on the Sun and their interaction with large scale

solar magnetic field, including models for eruption of Coronal Mass Ejections

(CME). These computers will also make it possible to routinely produce

coupled simulations of space environment from the surface of the sun, through

interplanetary space, into the tightly coupled magnetosphere - ionosphere

system. Furthermore, they will enable ensemble forecasting environments

including data assimilation and branch predictions to make probabilistic

forecasts about the severity of impacts from CMEs and other solar transients.

Petascale computing coupled with detailed observations and advanced physics

based numerical simulations will make detailed space weather forecasts

possible in the GOES-R era.