Mars Needs Better Storm Forecasts
NEW YORK — Solar storms unleashed by the sun’s activity may threaten future Mars colonies in addition to Earth. Now NASA is investing in a forecasting system that someday might give up to three days’ advance warning of space weather events.
Solar storms become more frequent during the peak activity periods of 11-year solar cycles. Earth has some protection because of its magnetic field and atmosphere, but Mars has only a very weak magnetic field and a thin atmosphere that would allow more radiation to reach its surface.
“Because Mars does not have the protections that Earth does, the significance of storms is much greater for a Mars colony,” said Roger Dube, a physicist at the Rochester Institute of Technology in New York.
Powerful space storms can send out a deadly wave of X-rays and high-energy particles that could have severe consequences for millions of people on Earth. One such solar flare knocked out the power supply in Quebec and disrupted power and communications across North America during a March 1989 solar storm.
“It has become more important to have space weather predictions because of the impact [the storms] can have on satellites, power grids, GPS and telecommunications,” Dube said.
The same threat would exist for lunar bases, or a manned mission on the way to Mars or elsewhere.
Building a better solar forecast means more than just launching additional sun-watching satellites. Dube and his colleagues want to train artificial intelligence (AI) to recognize the crucial signatures that herald an impending storm.
Past efforts have tried to find such solar storm signatures from data collected by a single spacecraft, Dube said. The new study will use neural networks to sift through data from all satellites and solar observatories.
“We’re developing the AI algorithms to allow them to absorb not only conventional time-based data, but also to allow the neural networks to absorb and digest image data,” Dube explained.
Being able to study solar images would allow the AI software to take advantage of hardware such as NASA’s Solar Dynamics Observatory, which can observe the sun with 10 times better resolution than high-definition television.
The AI neural network will rely upon a parallel processing network with 600 times more computing power than a high-end office computer that contains a 3-gigahertz processor. That would allow the AI to quickly crunch the data coming from the many different sun-watching sources.
Sensors or small solar observatories on Mars could also add to the instruments keeping an eye on the sun. But Mars colonists would also need a system that could alert them to seek shelter from any incoming solar storm, because such storms do not have audible or visual cues for people experiencing them.
A second stage of NASA’s new effort involves developing an all-clear sensor that can gauge how a storm is progressing on the ground and signal when the storm has died down.
“We’ll have to test it on the space station, perhaps on the Moon at some point if there’s a mission, and finally place it with a colony on Mars,” Dube said.
Sensors might also protect travelers on their way to Mars or other destinations throughout the solar system. An alert would cue astronauts to find shelter inside a shielded safe room aboard the spacecraft until the danger had passed.
The most reliable predictions for dangerous space weather currently come from a joint NASA and European Space Agency mission at the L1 Lagrange point — one of several locations where the gravitational forces of the sun and Earth perfectly balance out the orbital motion of a spacecraft.
The U.S.-European Solar and Heliospheric Observatory and other solar satellites act “almost like buoys that transmit half an hour advance warning to Earth when they detect a monster wave” from a solar storm, Dube said.
NASA and the U.S. National Oceanic and Atmospheric Administration then send out advance alerts to commercial airlines, power companies or satellite operators, based on the heads-up from the thin line of sun sentinels.
Still, boosting forecasting power to three days could prevent close calls and ensure better preparations on the ground. After all, half an hour does not necessarily represent much time for humans to respond.
“You don’t know when the storm is going to hit,” Dube said. “It might hit at 2 a.m. on a Sunday.”