Scientists create first detailed computer model of sunspots
Scientists at the National Center for Atmospheric Research built on existing simulation software to create the most comprehensive and detailed computer model to date of a sunspot, running on an NCAR supercomputer.
Scientists at the National Center for Atmospheric Research in Boulder, Colo., have developed the most comprehensive and detailed computer model to date of a sunspot, allowing researchers to simulate the activity of a solar storm.
Understanding sunspots is necessary to an understanding of the Earth’s atmospheric system, said Matthias Rempel, a research scientist at NCAR’s High Altitude Observatory. The large amounts of energy and ionized material produced by sunspots have an impact on the Earth’s weather and climate, can affect electromagnetic activity both on the surface of the Earth and in its magnetic field, and impact satellites in orbit.
“This is the first time we have a model of an entire sunspot,” said Rempel, lead author of a paper describing the work in the journal Science Express. “Our simulations will advance research into the inner workings of the sun as well as connections between solar output and Earth’s atmosphere.”
Solar activity typically runs in 11-year cycles, and the National Oceanic and Atmospheric Administration recently announced that the sun has entered a new cycle, which is expected to be milder than usual with fewer sunspots and fewer solar storms. The forecast, released by the NOAA’s Space Weather Prediction Center, is welcome news for the operators of the Earth’s electronic infrastructures and those who depend on them. But severe storms remain a threat to satellites, astronauts and our electronic infrastructure.
The sunspot simulation was run in December and January on NCAR’s Bluefire supercomputer, an IBM Power 575 Hydro-Cluster machine capable of 76 teraflops. It is among the top 25 most powerful supercomputers and was the first of a class of highly efficient machines when installed in May 2008. The supercomputer replaced three computers with a combined performance of 20 teraflops and is used primarily for atmospheric and weather forecast simulations. The project was supported by the National Science Foundation.
“If you had asked me two years ago if we could have done this,” Rempel probably would have said no, he said. But when he began work on refining the simulation code, “I had the feeling that the time was right to start something like this.”
The original code had been produced at the University of Chicago.
The code was used for magnetic hydrodynamics and then adapted to solar physics, Rempel said. He said he made some changes in the numerical scheme, implemented some additional science on energy transfer, sped up the input-output and improved the scale.
As a theoretical astrophysicist, “you have to do a lot of running computer models, and you have to be familiar with coding,” he said.
There were problems with the revised simulation crashing initially, but a year of refining the code produced a stable program that was ready for the Bluefire supercomputer. It comprises 1.8 billion grid points spaced 10 to 20 miles apart in a virtual three-dimensional model measuring about 31,000 miles by 62,000 miles by about 3,700 miles deep in the solar surface. Prior to this, it had not been possible to render an entire sunspot with that degree of detail, Rempel said. Resolution had to be sacrificed, or the model would have been restricted to a portion of a sunspot. But it was difficult to define boundaries on the sunspot without losing data.
The model runs from 100 to 150 times slower than an actual sunspot develops, Rempel said. To show one hour of development requires up to 150 hours of computer time on 500 processors.
The simulation was able to incorporate a number of phenomena observed over the past 100 years into a model of sunspot behavior, Rempel said. Because of more limited computing power, past models had more restrictions that were not able to explain many things that had been observed.
NEXT STORY: DOD weighs greater use of social media