Scientists Successfully Predict When and Where Dangerous Solar Storms Are Likely to Happen

From CERES-Science

Breakthrough forecasting method validated by surprise solar eruptions on the far side of the Sun

NASA: Six Solar Flares Erupted February 1-4, 2026

(source: NASA  https://x.com/nasagoddard/status/2019495176759124227?s=43; https://science.nasa.gov/blogs/solar-cycle-25/)

Mexico City, Mexico – February 14, 2026. A team of scientists from around the world has created the first system that can predict when and where extremely powerful solar storms, called superflares, are most likely to happen. These storms can disrupt power grids, communications, and satellites, and even pose dangers to astronauts in space.

Instead of trying to predict the exact moment a solar storm will erupt (which is nearly impossible), this new approach identifies extended windows of time—ranging from several months to a year—when the Sun is more likely to produce these extreme events. The method also pinpoints which regions of the Sun are most at risk. The research has been published in the Journal of Geophysical Research: Space Physics.

How the Forecasting System Works

The scientists analyzed nearly 50 years of data (1975–2025) from satellites that monitor the Sun’s X-ray emissions. They discovered two key patterns:

1. They identified specific zones on the Sun where magnetic energy builds up over time, making those areas more likely to produce powerful eruptions.

2. They found a rhythmic pattern in solar activity based on two natural cycles: a 1.7-year cycle and a 7-year cycle. When these cycles line up in certain ways, the risk of superflares increases significantly.

Using advanced mathematical techniques and machine learning, the team combined these patterns to forecast high-risk time periods and locations on the Sun. For the current solar cycle (Solar Cycle 25), their model predicts two main danger windows:

• Mid-2025 through mid-2026 (focused on the Sun’s southern hemisphere, between 5°S–25°S latitude)

• Early-to-mid 2027 (focused on the Sun’s northern hemisphere, between 10°N–30°N latitude)

Real-World Implications

Lead researcher Dr. Victor M. Velasco Herrera, from the National Autonomous University of Mexico, explained:

“Traditional solar forecasting struggles with these extreme events because they happen so quickly and unpredictably. Our method gives space weather operators and satellite managers one to two years of advance warning about when conditions are most dangerous. This critical lead time allows them to prepare and protect communications systems, power grids, and astronaut safety.”

Dr. Velasco Herrera also noted the relevance for space missions:

“NASA is right to postpone the Artemis II mission to the Moon until March, but given how active the Sun is right now, our forecasts suggest that delaying the launch until the end of 2026 may be a much safer decision.”

Surprise Validation from the Far Side of the Sun

The team got an unexpected confirmation of their model’s accuracy during the research process itself. After they submitted their paper for publication (between October and December 2025), other scientists using the Solar Orbiter spacecraft discovered a series of massive superflares that had erupted on the far side of the Sun—the side we can’t see from Earth. These eruptions included an X11.1 class on May 14, 2024, an X9.5 and X9.7 on May 15, 2024 and an X16.5 class on May 20, 2024, as reported coincidentally on the same day by a NASA study.

Remarkably, these newly discovered far-side events matched the patterns the team had predicted, even though they had no knowledge of these storms when developing their forecasting system.

“This was pure luck, but also very revealing,” said Dr. Velasco Herrera. “We created our forecast without knowing about these far-side superflares. When they were discovered during our paper review process, they aligned perfectly with our predicted patterns. This shows that our physics-based approach works across the entire Sun, not just the side facing Earth.”

Co-author Dr. Willie Soon, from the Center for Environmental Research and Earth Sciences (CERES), added:

“Nature gave us the perfect test. These far-side discoveries essentially validated our method in real time, proving that the underlying patterns we identified are reliable and work everywhere on the Sun’s surface.”

Why This Matters

Solar superflares are the most powerful eruptions the Sun can produce. A direct hit from one of these storms could cause widespread power outages, damage satellites, disrupt GPS navigation, interfere with radio communications, and create radiation hazards for astronauts and airline passengers at high altitudes.

By providing advance warning of when and where these events are most likely to occur, this new forecasting system gives utilities, satellite operators, and space agencies valuable time to take protective measures—such as adjusting satellite orbits, preparing backup systems, or rescheduling space missions.

As Solar Cycle 25 continues to show strong activity, this breakthrough offers a significant improvement in our ability to prepare for and mitigate the impacts of extreme space weather.

The research paper, titled ‘A New Method for Probabilistic Spatiotemporal Forecasts of Solar Soft X-ray S-Class (> X10) Superflares,’ is now published in the Journal of Geophysical Research: Space Physics (American Geophysical Union), https://doi.org/10.1029/2025JA034977.

The author team includes scientists from Mexico, the United States, Hungary, Turkey, Russia, Czech Republic, and China.

Publication Details

V. M. Velasco Herrera, G. Velasco Herrera, W. Soon, A. Özgüç, N. Babynets, A. Tlatov, M. Švanda, S. Qiu, S. Baliunas, B. Kotan, G. González González, L. A. Bautista Flores, M. Pazos (2026). “A New Method for Probabilistic Spatiotemporal Forecasts of Solar Soft X-Ray “S-Class” (>X10) Superflares“. Journal of Geophysical Research: Space Physics. 131 (2), e2025JA034977. https://doi.org/10.1029/2025JA034977

Get notified when a new post is published.
Subscribe today!
5 7 votes
Article Rating
24 Comments
Inline Feedbacks
View all comments
February 17, 2026 10:12 pm

Real, solid science. I was starting to miss that. I’m glad.

Eng_Ian
Reply to  Charles Armand
February 18, 2026 12:47 pm

I’m not so sure that this should be flagged as ‘Real solid science’. There are a lot of unstated details still required. I did note that the work is based on cycle lengths of 1.7 and 7 years duration, leading to a prediction of an enhanced risk with a duration of around 1 year, (see example from text).

I don’t want to sound too negative on this but what value does this add? As quoted, two windows were listed, one of 12 months, ending in in mid 2026, the next one starting in mid 2027, (no duration given). Is this a guaranteed 12 months clear between them or just a reduced probability?

Space programs are longer than these forecast duration periods, we don’t evacuate the ISS for a year at a time. In reality, a solar flare is an almost instantaneous event, it also has to be aligned to have any impact. I’d be more interested in a forecast that could tell the latitude of the source, if it is too far from the equator, (Sun’s), then the impacts will be reduced.

The article would have had more value if the probabilities of an event were listed for the periods of risk compared to periods of lull, for example is it a 10x difference of a 2x difference? Without that information a reader could think that no risk occurs between the windows and hence this is truly a valued insight. However, if the difference is akin to a 30% risk enhancement, then this is no better than a weather forecast for 2040, in summer it will be hot, and you still won’t know which day to carry an umbrella.

More info required.

Reply to  Eng_Ian
February 19, 2026 2:49 am

I quite agree, actually. But I’m so overwhelmed by catastrophic predictions and media opportunism that, when I see honest work that highlights progress and innovation without shouting about the end of the world, I tend to get enthusiastic. Maybe a bit too quickly, I admit, but it’s in my nature.

Scarecrow Repair
February 17, 2026 10:14 pm

The climate catastrophe community should be so good.

Neil Pryke
February 17, 2026 10:15 pm

And where does the human species fit into this system..?

Sparta Nova 4
Reply to  Neil Pryke
February 18, 2026 6:56 am

You need to provide a context for your question.

The basics are the human species lives and dies by the sun, but it is unclear what your comment addresses.

Alan
Reply to  Neil Pryke
February 18, 2026 9:04 am

It’s a system humans have no control over. All we can do is watch and learn.

leefor
February 17, 2026 10:37 pm

“ranging from several months to a year” Yep be very afraid it is gunna happen in the uncertain future.

Bryan A
Reply to  leefor
February 17, 2026 10:51 pm

I was wondering the same thing…let’s see how good I do…
There will be an X class flare in the northern equatorial region that will throw a CME towards Earth making the Aurora visible in Northern CA through Northern Texas lasting 2 days.

Sparta Nova 4
Reply to  leefor
February 18, 2026 6:57 am

Sort of like I predict there will be a hurricane in the Atlantic in 2026.

Art Slartibartfast
February 17, 2026 11:15 pm

“After they submitted their paper for publication (between October and December 2025)”…well, that they cannot exactly pinpoint the date of the eruption is no surprise if they cannot exactly say on which date they submitted their paper 🙂

Jokes aside, that is some good scientific work. As measurements build the probability distributions of eruption times will become more accurate and maybe the conclusion will be that the distributions remain broad.

Sparta Nova 4
Reply to  Art Slartibartfast
February 18, 2026 6:57 am

Sort of like getting to the cyclical nature of climate variations would be good.

strativarius
February 18, 2026 12:57 am

Months to a year…

Not 2100, then

strativarius
February 18, 2026 1:44 am

Story tip:

REPORT: Labour’s Ban on North Sea Oil Drives Up Carbon Emissions
https://order-order.com/2026/02/18/report-labours-ban-on-north-sea-oil-drives-up-carbon-emissions/

Californian oil instead?

Bruce Cobb
February 18, 2026 2:37 am

OK, the Sun needs to stop eating beans and then lighting off the result. It’s not funny any more.

MarkW
February 18, 2026 6:07 am

The article mentions one successful prediction.
That’s great.
But I wouldn’t start celebrating until they have a string of at least 10 successful (or at least near successful) predictions.

Eng_Ian
Reply to  MarkW
February 18, 2026 12:54 pm

I don’t think that a single event, even ten is a success. This can only be considered successful IF it predicts a high risk window, with a specific rise above the norm, eg 10 major storms AND also predicts the lull, eg 1 major storm.

The dates of the highs and the lulls must align and the magnitude of the signal must also match, anything less could be interpreted as an non falsifiable scientific prediction.

Surely the paper listed the dates AND the magnitudes/number of major storms for their windows…. AND compared that to a forecast for the period between the peaks. Or did they?

Sparta Nova 4
February 18, 2026 6:54 am

One unexpected flare that falls within the prediction window does not validate the whole thing.

It could, of course, but we will need more successes before validation can be claimed.

That they are not predicting amplitude or specific times is proper and giving hazard warninga based on probabilities is of value. Let us hope there are not a significant number of false alarms that would shelf what could be a useful tool.

drh
February 18, 2026 7:13 am

Story tip: https://finance.yahoo.com/news/gasoline-starved-california-turning-fuel-160000840.html

California is importing more than 40% of its refined petroleum from the Bahamas. Not very climate friendly when local refineries are shuttered only to require a 5000 mile transit for tankers to bring the fuel. How long before gasoline in CA is over $8 per gallon?

Harry Durham
February 18, 2026 7:25 am

Based just on history (i.e., no data other than ‘xxit happens’) I hope they continue their work until they decipher the next – or next two, or three – level cycle(s) in the periodicity.

To go along with several other posts with predictions, here’s mine:

The more inherent cycles they discern, the better the accuracy of their prediction.

Of course, since I’m in my eighth decade, I doubt I’ll be around to confirm my prediction…

Pop Piasa
February 18, 2026 8:09 am

I wish Lief Svalgaard could still be around to weigh in on this story. May his memory live on at WUWT.

February 18, 2026 9:41 am

This paper is not a deduction from 1st principles, but a phenomenological correlation. Nothing wrong with that but it is not a prediction. AI is being used to find ‘probabilities’ because it can easily search and correlate very large data sets.
It is in much the same vein as El Nino ‘predictions’ – watching for the southern oscillation in pressure to begin making a probabilistic prediction. ‘Plant rice this year’ as it probably will rain more than usual. Useful, but not earth-shaking, and quite possibly wrong. The 2016 El Nino known affectionately as ‘Godzilla’, had more La Nina characteristics than El Nino. For example, the probability was for lots of rain to end CA’s drought. Didn’t happen. Such ‘science’ can create a state of fear, and maybe that is the point. ‘Watch out for Storm Fern. It is a doozy.’ People love to be frightened into doing dumb things.

Eng_Ian
Reply to  whsmith@wustl.edu
February 18, 2026 1:02 pm

It’s also based on a short sample duration. The whole pattern could be thrown out with just another couple of data points, (eg 12 months more data).

It’s a little like making a forecast on the cycle length of even numbers at a roulette wheel when all you have is the last 10 spins to evaluate.

In reality, there may be no true signal at all, just a desire to find something rhythmic amongst the noise.

Zeke
February 18, 2026 5:37 pm

Now first, considering that solar activity has exceeded the other forecasts by a lot, isn’t this introducing a new forecast taking said increased solar activity into account?

And second, considering that humans have been watching the Sun and keeping sunspot records for a little while, we know that the sunspots will intensify in the northern solar hemisphere and also in the southern hemisphere, at offset times.

I just don’t want to hear about it when some TechBro takes a victory lap for his exponentially expensive model and claims ai taught us everything we know about the sun.

This is not modeling, or prediction, this is historical revisionism. Like I said, I don’t want to hear it.

It’s probably really about private equity (debt) being marketed to retirement accounts to fund datacenters. That’s what this is about isn’t it…grumbles, walks off…