ith his eye on the the sun and the wind, Charles Goodrich at the University of Maryland is watching out for storms. But the weather he's studying extends for millions of miles around our planet, and the geomagnetic storms he's trying to predict are caused by huge streams of interplanetary plasma interacting with the Earth's magnetic field and upper atmosphere.
Gusts of solar wind drive changes in the Earth's magnetosphere and ionosphere, with consequences both impressive and dangerous. One indication of a geomagnetic storm is an intense auroral display, caused when high-energy particles trapped by our planet's magnetic field excite atoms of oxygen, nitrogen, and argon in the ionosphere (Figure 1). But electromagnetic surges from the storms also can disrupt radio communications and cause blackouts in power grids. Even the smallest geomagnetic events, the "substorms" that can occur several times a day, deliver several megatons of energy--the equivalent of a large nuclear explosion--to the ionosphere in only a few minutes.
Figure 1: The Aurora from Above
This POLAR satellite X-ray view of the arctic during the event of January 1997 shows the auroral oval, the region in which the ionosphere interacts with charged particles trapped in the Earth's magnetic field and emits visible light and long- and short-wavelength electromagnetic radiation.
The research of the Space and Plasma Physics group, led by Dennis Papadopoulos at the University of Maryland, focuses on the physics of space plasmas, ranging from high-altitude lightning to shocks in supernovas. One of the group's major objectives is to explore the interactions between the Sun, the Earth, and interplanetary plasma, and to predict the state of the geospace environment. Satellites and observatories gather the data, but analyzing the information and modeling the processes that operate in the magnetosphere requires the power of a supercomputer.
"Numerical modeling is crucial to our understanding of space plasmas," Goodrich said. "We use a range of different simulation codes to model different phenomena. Most of our applications can run well on massively parallel systems like the CRAY T3E, but we plan to use the T90 for post processing--graphics in particular--and for developing enhancements to our algorithms and models." Goodrich has long been a user of computing resources at SDSC and is a former member of the SDSC Steering Committee. The Space and Plasma Physics group is adapting to the end of the NSF's Supercomputer Centers program by closing out its other accounts at Pittsburgh and Cornell and concentrating its heavy computation at SDSC under NPACI.
Goodrich and his colleagues Ray Lopez, graduate student Michael Wiltberger, and John Lyon of Dartmouth College are using magnetohydrodynamic (MHD) simulation codes to model space plasma physics. A fully 3-D MHD model numerically simulates the dynamics of the magnetosphere and can derive current magnetospheric conditions from spacecraft measurements. The researchers then compare the results of the simulations directly against data from spacecraft and ground observations.
Simulation speed and quality depend on the number of computational cells used to model the volume around the Earth. Until recently, the simulations used grids of about 40,000 cells, modeling the ionosphere in chunks three or four hundred kilometers across at a cost of 15 T3E processor-hours per hour of simulated time. To model the finer structures that give rise to ionospheric convection, the researchers will use more than 500,000 cells, to get 100-kilometer resolution at a cost of 300 T3E processor-hours per simulated hour.
Recent results have revealed that the terrestrial magnetosphere acts like a lens, focusing energy carried by the solar wind to a region behind the Earth where major geomagnetic occurrences originate (Figure 2). The electromagnetic energy carried or created by the impact of the solar wind on the Earth's magnetosphere concentrates at a distance of about 10 Earth radii (40,000 miles), leading to loss of the delicate magnetospheric equilibrium inducing a substorm. The conditions under which magnetospheric focusing occurs depends critically on the orientation of the magnetic field in the solar wind. The researchers expect that new data from the fleet of satellites probing the magnetosphere in the International Solar Terrestrial Program (ISTP) will further reveal how the magnetosphere can act like an enormous magnifying glass.
"For years people have puzzled over observations indicating that substorms initiate on the far side of the Earth from the sun, about 10 radii out," Goodrich said. "Identifying this focal point is equivalent to finding the 'smoking gun' that causes substorm phenomena."
Figure 2: Geomagnetic Storm
In January 1997, a dense magnetic plasma cloud crashed into the Earth's magnetosphere. The impact was modeled in a magnetohydrodynamics simulation by the Space Plasma Physics group at the University of Maryland. Thisfrom a 10-minute visualization was created with the assistance of Maryland's Advanced Visualization Laboratory; portions of the video are available on the Web. The color scale shows density, ranging from high (red) to low (blue). The simulation showed a low-density "hole" behind the shock front (yellow).
THE GREAT STORM OF JANUARY 1997
In early January of last year a dense cloud of plasma was ejected from the Sun on a collision course with our planet. From its birth, in a solar eruption detected by the NASA-ESA Solar and Heliospheric Observatory (SOHO) satellite, the blob was intensively monitored by ground-based observers and a flotilla of spacecraft. When it reached the Earth, the cloud's own magnetic field and the blob of plasma trailing 30 million miles behind it first caused intense geomagnetic activity and then compressed the front of the magnetosphere.
As the cloud engulfed Earth, NASA's WIND and GEOTAIL spacecraft measured the solar wind speed, the magnetic field strength and direction, and the plasma density. The event was followed through the magnetosphere to the ionosphere and the ground; auroral activity was observed from the ground and by X-ray imagers on the POLAR satellites and was correlated with ground-based magnetometer observations.
At the height of the event the electrical power dissipated in the aurora was approximately 1,400 gigawatts, nearly double the electrical power generating capacity of North America. By the time the storm was over, AT&T's Telstar 401 commercial communications satellite was a casualty of the electromagnetic effects.
Natural physical systems often have nonlinear and dissipative properties, and exhibit complex behavior that is difficult to understand and predict. But these nonlinear dissipative systems often can be described in terms of a few variables; the complexity arises from sensitivity to initial conditions. Analysis of observational time series data of the solar wind and the magnetosphere reveals that a relatively small number of parameters can describe the system. Researchers are learning to use the techniques of nonlinear dynamical systems to predict the occurrence and intensity of magnetic storms and substorms, creating a forecasting tool for space weather.
As part of the ISTP study, the Space and Plasma Physics group performed an initial global MHD simulation of the encounter using Pittsburgh Supercomputing Center resources. The unprecedented wealth of data allowed them to model the entire event from the arrival of the shock front preceding the cloud through the next 36 hours of the geomagnetic storm. From their simulation they produced a time-lapse video of the event. Their research continues with NPACI resources.
"NASA needs to acquire the capability to accurately predict the state of the geospace environment," said Goodrich, whose group's solar-terrestrial physics effort is sponsored primarily by NASA's Global Geospace Science project. "The January 1997 event showed that a violent space storm can destroy a spacecraft, so this information could turn out to be vital to the success of future missions." --MG