Earth’s magnetosphere might not protect us from superstorms after all

Science seems like a straightforward endeavor. You come up with a hypothesis, collect data to prove or disprove it, and analyze that data to see whether the hypothesis is right. But anyone who actually does science will tell you that it is often not that straightforward. One of the most common complexities is data analysis.

A new paper in Nature suggests that one such complexity, known as regression to the mean, might be causing us to massively underestimate how severe solar storms can be.

The interaction between the solar wind and Earth’s magnetosphere can be thought of as a giant dynamo. The combination of solar wind and magnetic field drives high-speed plasma and electric currents into the upper atmosphere, particularly around the polar caps (hence the appearance of auroras). Scientists typically track this dynamo effect using the Polar Cap Index (PCI).

One thing has been clear for decades: For moderate solar activity, there is a clear linear relationship between solar wind electric fields and the electric response measured on Earth. However, for stronger storms, that relationship breaks down as Earth’s response seems to “saturate”—that is, cap out at a certain level below what would otherwise be expected. But we never really understood why.

According to the new paper by Dr. Nithin Sivadas and their co-authors at NASA’s Goddard Space Flight Center, that saturation might simply be an illusion caused by how we measure the strength of solar storms. They noted that most solar wind measurements come from satellites like WIND, ACE or DSCOVR, all of which are located at the L1 Earth–Sun Lagrange point—which means they are 1.5 million km (930,000 miles) closer to the sun than Earth is.

That distance introduces a great deal of uncertainty. The timing of wind propagation varies, making precise timing difficult. The wind itself can change in the intervening 1.5 million km (930,000 miles), and shock fronts can introduce “heteroskedastic noise”—that is, random errors that grow larger in extreme events.

Those errors create a situation in which we are measuring a highly uncertain value far upstream, causing a fundamental rule of statistics to come into play: regression to the mean.

Essentially, scientists paired extreme measurements at the L1 satellites with the smaller, more average geomagnetic responses they triggered in the magnetosphere, since the “true” solar wind hitting Earth’s magnetosphere is much more likely to be less extreme and closer to the mean measured there.

This mathematical mismatch—of an extreme event at L1 and a more moderate one at Earth—causes a “nonlinear regression bias” in the data curve, artificially bending it and making Earth’s response appear “saturated.”

To prove their point, the authors applied a “regression calibration” to offset some of the bias. After doing so, the linear relationship between solar storm strength and Earth’s magnetic response continued with no clear saturation effect. Put simply, a 1-in-1,000-year solar storm event would now be much more likely to cause massive destruction than we originally thought.

But this finding is not relevant only to space weather. The authors point out that everything from seismology to medical trials can fall victim to the mean masquerading as a threshold. Machine learning exacerbates the problem, as models trained on uncertain inputs will simply learn the statistical illusion and think it is a physical reality.

The immediate takeaway is clear, though: Operators of Earth’s power grids and satellite constellations need to consider the possibility of a much larger electrical spike if a severe solar storm hits. The natural protection we thought we had from the magnetosphere’s saturation unfortunately appears to be just a statistical illusion.

Who’s behind this story?

Sadie Harley

BSc Life Sciences & Ecology. Microbiology lab background with pharmaceutical news experience in oil, gas, and renewable industries.

Full profile →

Andrew Zinin

Master’s in physics with research experience. Long-time science news enthusiast. Plays key role in Science X’s editorial success.

Full profile →