Major Solar Flares Raise Northern Lights Probability Across 23 US States

On June 3, 2026, Active Region 4455 on the sun's surface erupted three times in rapid succession: an M9.3 flare at 01:36 UTC, an M7.7 at 07:00 UTC, and an X1.0 at 11:28 UTC [1][2]. The X-class flare — the most powerful category in solar classification — triggered R3 (strong) radio blackout conditions across the Americas and sent a coronal mass ejection (CME) racing toward Earth [2]. That faster CME overtook the earlier one from the M7.7 flare, merging into what forecasters call a "cannibal CME" — a combined solar wind structure expected to strike Earth's magnetosphere late on June 4 and persist through June 5 [1][2].

NOAA's Space Weather Prediction Center (SWPC) responded with a G3 (Strong) Geomagnetic Storm Watch for June 4-5, with the possibility of G4 (Severe) conditions [3][4]. The agency's aurora forecast maps show the northern lights potentially visible across 23 U.S. states, drawing widespread media coverage and public excitement.

But the spectacle of green and purple skies obscures a harder set of questions: about the fragility of infrastructure that depends on a quiet magnetosphere, about who actually receives these warnings and who doesn't, and about whether the forecasts themselves can be trusted.

The Flares in Context: Powerful, but Not Historic

Solar flares are classified on a logarithmic scale. An X1.0 flare — the designation for the June 3 event — sits at the threshold of the X-class category, which begins at X1.0 and has no upper bound [2]. To calibrate expectations: the May 2024 Gannon Storm, which produced visible auroras as far south as the Florida Keys and Puerto Rico, was driven by an X8.7 flare [5]. The Halloween storms of October 2003 peaked at an estimated X35, and the Carrington Event of 1859 — the most intense geomagnetic storm in recorded history — is estimated at roughly X45 [6][7].

Source: NASA / NOAA SWPC

Data as of Jun 5, 2026CSV

In other words, the June 2026 event is roughly 45 times less intense than the Carrington Event and 8.7 times weaker than the Gannon Storm by raw X-class rating. What makes it significant is not raw power but geometry and timing: the CMEs are Earth-directed, they are merging en route, and they arrive during a period when Solar Cycle 25 continues to produce elevated activity despite having likely passed its peak in October 2024 [8][9].

Which 23 States, and What Does "Severe" Actually Mean?

The NOAA geomagnetic storm scale runs from G1 (Minor) through G5 (Extreme), corresponding to Kp index values of 5 through 9. A G3 storm corresponds to Kp 7; G4 corresponds to Kp 8-9 [3]. At these levels, the equatorward boundary of the auroral oval — the ring of atmosphere where charged solar particles excite atmospheric gases — pushes significantly south.

Source: NOAA SWPC

Data as of Jun 5, 2026CSV

At G3 conditions, NOAA's models show the aurora potentially visible down to approximately 50° geomagnetic latitude. At G4, that boundary drops to roughly 45° [3]. The 23 states under the aurora alert span a wide band across the northern and central U.S.: Washington, Oregon, Idaho, Montana, Wyoming, North Dakota, South Dakota, Minnesota, Wisconsin, Michigan, Iowa, Nebraska, Illinois, Indiana, Ohio, Pennsylvania, New York, Vermont, New Hampshire, Massachusetts, Connecticut, Rhode Island, and Maine [10][11].

A critical distinction: "potentially visible" on NOAA's probabilistic models does not mean a bright overhead display. For most of the southern states on that list — Illinois, Ohio, Connecticut — any aurora would likely appear as a faint glow on the northern horizon, visible only under dark skies far from city light pollution, during the 10 p.m. to 2 a.m. peak window [10]. During the far more powerful Gannon Storm in May 2024, citizen science reports showed that NOAA's OVATION Prime model actually underestimated how far south the aurora reached at extreme Kp levels [12]. But an X1.0-driven G3/G4 storm is not a G5 event, and expectations should be calibrated accordingly.

The Infrastructure Nobody Sees

While social media fills with aurora photography tips, utility operators across the affected states face a different calculus. Geomagnetic storms induce electric fields at ground level through Faraday's law, driving quasi-DC geomagnetically induced currents (GICs) through transmission lines, pipelines, and transformer neutral connections [13][14]. High-voltage transformers are particularly susceptible: GICs cause half-cycle core saturation, generating harmonics, increasing reactive power consumption, and producing localized heating that can degrade insulation or, in severe cases, cause catastrophic failure [13].

The precedent is concrete. On March 13, 1989, a severe geomagnetic storm collapsed the Hydro-Québec power grid in seconds as protective relays tripped in a cascading sequence, cutting power to six million people for nine hours [13]. During the Halloween storms of 2003, twelve transformers in South Africa were taken out of service [6].

The U.S. grid's vulnerability is structural. Large power transformers (LPTs) are the backbone of high-voltage transmission — and they are few in number, enormous, and functionally irreplaceable on short timelines [15][16]. Current lead times for large power transformers range from 115 to 210 weeks — roughly two to four years [16]. A single unit costs between $1.5 million and $4 million depending on capacity, and prices have surged 60-80% since 2020 [16]. Over 80% of LPTs installed in the U.S. are imported [15]. A GAO report has flagged that if even a handful of these transformers were disabled simultaneously — whether by geomagnetic storm, sabotage, or other cause — entire regions could lose power for months [15].

Neutral blocking capacitors, a technology designed to prevent GICs from entering transformer windings, have been developed and operationally tested but have not been widely deployed across the grid [13]. For a G3/G4 event like the current forecast, the risk of transformer damage is lower than it would be during a G5 storm, but it is not zero — and the replacement pipeline cannot absorb even modest losses quickly.

Source: OpenAlex

Data as of Jan 1, 2026CSV

Academic interest in this vulnerability has grown sharply: over 4,000 papers on geomagnetic storms and power grid impacts have been published since 2011, with a peak of 520 in 2025 alone [17].

GPS, Farming, and the Economic Exposure Window

The economic footprint of a geomagnetic storm extends well beyond the power grid. GPS signals, which rely on precise timing from satellites in medium Earth orbit, are degraded by ionospheric disturbances during geomagnetic activity [18][19]. Precision agriculture, autonomous vehicles, aviation navigation, financial trading systems, and emergency services all depend on GPS accuracy [18].

The May 2024 Gannon Storm provided a real-world cost estimate: GPS disruptions during a critical spring planting window cost American farmers over $500 million in potential profit, as precision-guided tractors veered off programmed routes [18]. Airlines were forced to reroute polar flights, adding fuel costs and delays [18]. A broader estimate from risk modeling puts the total first-year economic damage from a severe geomagnetic storm at $0.6 to $2.6 trillion for the United States [18][20].

For the current G3/G4 forecast, the economic exposure is more modest but still real. HF radio communications — used by transoceanic aviation, maritime operations, and military systems — experienced R3-level blackouts during the X1.0 flare on June 3, and further degradation is expected as the CME arrives [2]. Satellite operators typically enter protective modes during G3+ storms, temporarily reducing service availability [19].

Can You Trust the Forecast?

NOAA's aurora forecasts operate on two timescales. The OVATION Prime model provides a 30-to-90-minute forecast based on real-time solar wind measurements from the L1 Lagrange point, about 1.5 million kilometers sunward of Earth [21]. At that short range, accuracy is reasonable — the solar wind data is direct and the travel time to Earth is known.

But the multi-day forecasts that drive the "23 states" headlines rely on CME propagation models like ENLIL, which predict when and how forcefully a CME will hit Earth's magnetosphere. These models carry significant uncertainty. A comprehensive 26-year verification study of SWPC solar flare forecasts found that they are "poorly calibrated and produce excessive false alarms, particularly for X-class flares" [22]. The study concluded that the SWPC model "does not outperform baseline forecasts" and exhibits severe calibration issues in high-stakes scenarios [22].

This matters for trust. When NOAA issues a "severe" storm watch and the aurora fails to materialize at the predicted latitudes — as happens with some regularity — the false alarm erodes confidence among both the public and the utility operators who must decide whether to take costly protective measures like reducing transmission loads or activating blocking devices. Conversely, the Gannon Storm showed that extreme events can exceed model predictions, catching systems off guard [12].

The forecast skill gap is a known problem within the space weather community, but closing it requires better solar wind measurements, improved CME modeling, and a monitoring infrastructure that has not received commensurate federal investment.

Solar Cycle 25: Declining but Still Active

The timing of this event reflects where we stand in the sun's 11-year activity cycle. Solar Cycle 25 likely reached its peak — or solar maximum — in October 2024, with monthly sunspot numbers averaging 143.5 at peak [8][9]. That figure makes Cycle 25 significantly more active than the weak Cycle 24 (peak: 81.9 sunspots) and closer to Cycle 23 (120.8) and Cycle 22 (158.5) [8].

Source: NOAA SWPC / WDC-SILSO

Data as of Jun 1, 2026CSV

Even in the declining phase, intense individual events continue. Since January 2026, Active Region 4366 alone has produced 21 C-class flares, 38 M-class flares, and six X-class flares, including an X8.3 on February 2 [9]. The pattern is consistent with historical cycles: some of the most consequential geomagnetic storms have occurred during the declining phase, when large, stable sunspot regions persist and coronal holes become more prevalent.

Solar Cycle 25 has also exceeded early predictions. NOAA's initial 2019 forecast called for a cycle "similar to Cycle 24" — below average [23]. Instead, Cycle 25 has run roughly 31% above Cycle 24's pace through its first several years, a gap that revised forecasts now acknowledge [9].

The Aurora-Chasing Paradox

Every "severe" aurora alert generates a wave of public activity: social media excitement, late-night drives to dark-sky locations, and a surge in what the travel industry calls "noctourism" [24][25]. The educational value is real — space weather events are among the few moments when solar physics becomes tangible to millions of people.

But the steelman case for net harm is worth stating. Aurora chasers driving rural roads late at night in unfamiliar areas face elevated accident risk. Trespassing on private land to reach dark-sky vantage points is a recurring complaint from rural property owners during alert windows. And the growing phenomenon of "aurora tourism" is placing pressure on fragile ecosystems, particularly in Arctic and sub-Arctic regions where increased visitor traffic coincides with sensitive wildlife periods [25].

For the continental U.S. events like this week's, the ecological impact is minimal. The safety question is harder to dismiss — though no systematic data exists on aurora-related traffic incidents, which itself reflects how under-studied the phenomenon is. The counterargument is straightforward: the risk profile of aurora chasing is comparable to any nighttime recreational driving, and the public engagement with science during these events has durable educational value that outweighs marginal safety costs.

Who Doesn't Get the Warning

NOAA's SWPC communicates primarily through its website, email subscription lists, and social media channels. The National Weather Service amplifies space weather alerts through its standard broadcast infrastructure. But this pipeline has known gaps.

An estimated 39% of rural Americans lack broadband internet access [26]. NOAA's space weather products are published exclusively in English, with no systematic translation into Spanish or other languages spoken by significant U.S. populations. For the roughly 42 million Spanish speakers in the U.S., SWPC alerts are functionally inaccessible unless filtered through bilingual media coverage.

The equity dimension extends to infrastructure dependence. People in rural and low-income communities who rely on GPS-enabled medical devices — insulin pumps with location services, personal emergency response systems, GPS-tracked medication deliveries — face disruptions they may not understand or anticipate. NOAA's public hazard communication mandate does not currently include targeted outreach to these populations for space weather events, a gap that the agency's own modernization efforts have not yet addressed [4].

This is not unique to space weather — it mirrors broader gaps in the National Weather Service's communication infrastructure for all hazard types. But the relative obscurity of space weather as a hazard category means that affected populations are even less likely to have prior awareness or adaptive capacity.

What to Watch For

The CME complex is expected to arrive at Earth late on June 4 or early June 5. NOAA forecasts G3 conditions with the possibility of G4 [3][4]. By June 6, activity is expected to drop below G1 [3].

For aurora watchers in the 23 affected states: the best odds are between 10 p.m. and 2 a.m. local time on the night of June 5, under clear skies away from urban light pollution. Northern-tier states — Montana, Minnesota, Michigan, Maine — have the best chances of a visible display. For states at the southern edge of the alert zone, manage expectations: any aurora will likely be faint and low on the northern horizon.

For utility operators, satellite companies, and GPS-dependent industries: the storm represents a manageable but non-trivial disruption window. The real test will come when — not if — a G5-class event arrives during this solar cycle's extended active period, and the grid's aging transformer fleet faces currents it was never designed to carry.