Fact-Checked Science

Guide · Earth’s magnetic field

Earth’s magnetic field is weakening slowly; storms, not a flip, are the near-term risk

Earth’s magnetic field is weakening: its main part, the axial dipole, is about 9% weaker than in 1840 [1 p. 1]. This guide brings together our checks on the field, its past, its weak spot over the South Atlantic and space weather, and adds a full IGRF-14 analysis of the decline since 1900, set against two million years of records [2][3]. It covers how long the field’s big changes take, what a weaker field means for life, and the solar storms that are the real near-term risk.

By Fact-Checked Science · Published

Our analysis marks our own reasoning and calculations, including our runs of the IGRF-14, WMM2025, PADM2M and GGF100k model files and our counts in GFZ’s Kp and the Oulu neutron monitor’s records, and the verdicts are our assessment: four repeat the site’s existing checks, the first states what its pages have always said, and the sixth is new. Everything else comes from the numbered sources. For the S0 News episodes and the video, times are positions in the video; for papers and reports, p. gives a page, § a section, Table a table and Abstract or another heading the part we read; for the UNSCEAR annexes, ¶ gives a paragraph.

A checklist for claimsCode and numbersDownload as PDFSourcesHow this was checkedCorrections

The field is weakening slowly; the scarier claims about it fail or overstate the risk

Our assessment

About 9%

how much weaker Earth’s axial dipole, the main part of the field, is than in 1840 [1 p. 1]

5% a century

on average since 1840; about 6% since 1900; from one five-year span to the next, between about 3% and 10% (our analysis of IGRF-14) [1 p. 2][2]

773,000

years since the last full reversal of the field, the Matuyama–Brunhes [4 Teaser]

The common claims

True

The field is weakening

“the weakening magnetic field of Earth” (S0 News, October 24, 2026) [5 00:01:51–00:01:53]; “Earth’s magnetic field, that shield blocking all of the particle energy is weakening once again” (the video) [6 30:03]

It is, slowly. The axial dipole is about 9% weaker than in 1840 [1 p. 1]. In IGRF-14 the full dipole moment fell 7.6% from 1900 to 2025, about 6% a century, and is forecast to keep falling (our analysis) [2][7 p. 17]. Only this premise is rated here. The same October 24 passage offers the weak field as the explanation for the ozone hole, rated False on our review of the episode, and goes on to the extra exposure, which is claim 6; the video’s “15% gone, ten times faster” is rated at stop 6. See section 2.

False

Excursions come about every 6,000 years

“This is happening approximately every 6,000 years.” … “We’re that exact distance from the last one … We’re perfectly due in time.” (the video, stop 3) [6 28:45][6 29:45]

The record keeps no such rhythm. In our test of the cycle, a real 6,000-year cycle counted from today would leave the six best-dated excursions about 470 years from its marks; they sit 1,640 years away on average, which no simulated cycle produced in 200,000 tries. Reviews list 12 to 17 excursions since the last reversal, about 7 of them adequately recorded [8], with gaps in the last 120,000 years from about 5,000 to about 30,000 years (our review of the record). See section 1.

False

An excursion is under way now

“Luckily, the excursion that we’re in now isn’t that bad” (S0 News, October 22) [9 00:03:38–00:03:41]; “the geomagnetic excursions like we’re in now happen quickly” (October 23) [10 00:02:42–00:02:48]

Nothing measured is near one. Excursions are commonly marked by a pole straying more than 45° from the geographic pole, with the dipole at about half today’s strength [11 §4.4]. Today the dipole axis is tilted 9.2°, down from 11.4° in 1900, and the dipole is about 40% stronger than its average over the last two million years (our analysis) [2][3]. See sections 1 and 2.

False

Excursions take out much of life

“but it still takes out a large chunk of life every time it happens” (S0 News, October 22) [9 00:03:41–00:03:45]

A 2010 review finds “no clear picture of a geomagnetic link” between the field’s changes and the evolution of life [12 Abstract], and “reversal patterns do not match patterns in species extinction during geological history” [13]. The one argued case, the Laschamp, is contested (our review of the October 22 episode). At the ground the air does most of the shielding: even with no field at all, the cosmic-ray dose at sea level at low latitudes would rise by at most about a tenth of a millisievert a year, against about 3 mSv from all natural sources (our analysis) [14 ¶20–25][15 ¶284]. See section 5.

False

Earth turns over 90° in a day

“in one day during the magnetic pole shift, Earth turns over 90°” (the video, stop 17) [6 1:22:26]; alluded to on October 24: “the turning of the earth, swaying like a drunkard” [5 00:03:39–00:03:42]

Physics rules it out. Turning a 100-km-thick lithosphere 90° in a day takes about 7 × 10²⁶ J of motion within the day (our review of Earth’s energy stores), while all the heat leaving Earth comes to about 4 × 10¹⁸ J a day [16 Abstract]. Palaeomagnetic data limit such polar wander to at most about 3° per million years over the last 539 million years [17 Introduction]; 90° in a day is about ten billion times faster. See section 6.

Overstated

More vulnerable to solar storms

“Earth is letting in more space particle energy as our shield weakens” [5 00:01:57–00:02:02]; “Being more vulnerable to the sun and geomagnetic storms sounds extra scary” (S0 News, October 24) [5 00:02:03–00:02:07]

This rates the passage’s second half. The direction is right, the size is not: since 1900 the weaker dipole has cut the average cutoff cosmic rays must beat by about 8% [2], raised their production in the air by about 4% [18] and moved the auroral oval’s edge under a degree toward the equator [19 p. 3669] (our analysis). One storm moved that edge 20–35° in May 2024 (our check); production swings about ×1.8 over a solar cycle; NOAA’s radiation scale spans a factor of 10⁴ [20]. And over eastern and central North America the tilting axis has moved cities away from the auroral zone since 1900 [21 §1]. See section 7.

1 · The record

Excursions and reversals are real, irregular and slow

The field has kept its present polarity since the Matuyama–Brunhes reversal, whose final flip came 773,000 years ago [4 Teaser]. Since then it has made a dozen or more excursions, brief swings far from its usual axis that end back in the same polarity [8]; over the last 25 million years it has reversed about five times every million years [22 §5], at irregular intervals [13].

Today’s axial dipole against the long record

  • The model’s axial dipole moment
  • Today’s, from IGRF-14
  • The model’s average over the span
  • Below about half today’s

The last two million years, in PADM2M

0246810×10²² A m²Matuyama–Brunhes reversal, 773,000 years agoToday: 7.59Two-million-year average: 5.32Below about half today’s: the usual mark of an excursion2.01.51.00.5today

Million years ago

The last 100,000 years, in GGF100k

0246810×10²² A m²Laschamp excursion, about 41,000 years ago: 2.44Today: 7.59100,000-year average: 6.78Below about half today’s: the usual mark of an excursion10080604020today

Thousand years ago

Our analysis of the published model files: PADM2M every 2,000 years [3] and GGF100k every 250 years [23], with today’s value from IGRF-14 [2]. The shading marks an axial dipole below about half today’s, 4 ×10²² A m², the level commonly used to mark an excursion [11 §4.4]. Sediment-based models smooth fast changes, so the dips are softened [24 Abstract]: GGF100k’s Laschamp low of 2.44 is several times the lows, below 0.4 and even below 0.2, of the models built for the excursion [11 §4.2].

Today against two million years

Our analysis

1.43×

Today’s axial dipole, 7.59 ×10²² A m², against PADM2M’s average for the last two million years, 5.32 [2][3]. About 5% of that time had a dipole as strong.

Today against 100,000 years

Our analysis

1.12×

Against GGF100k’s average for the last 100,000 years, 6.78 ×10²² A m² [23]; four reconstructions put that average at 6.04 to 7.13 [24 §4.2].

Reversals per million years, recently

About 5

“The 5 Myr average reversal frequency has fluctuated about ~5 events per million years since about 25 Ma” [22 §5]. Reversals “are not predictable and are certainly not periodic in nature” [13].

An excursion is not a reversal

Excursions are “brief directional aberrations of the main dipole field outside the range of expected secular variation”, and the better-dated ones “imply millennial-scale or even sub-millennial-scale durations” [8]. In a reversal the field settles into the opposite polarity, as it last did 773,000 years ago, after excursions at about 795,000 and 784,000 years ago [4 Teaser].

What the records can and can’t resolve

Lake and seafloor sediments smooth the field as they lock it in. GGF100k builds that smoothing into its fit and has “less temporal complexity” than current models of the Holocene [24 Abstract]; global models of the Laschamp “represent smoothed field behavior. Rapid field changes may be distorted” [11 §5]. PADM2M resolves about ten thousand years [25 p. 4], and models like these “do not include formal uncertainty estimates” [25 p. 14].

Our analysis

Coming down from a high

In PADM2M the axial dipole was at or above today’s value for about 5% of the last two million years and 13% of the last 780,000; in GGF100k, for 25% of the last 100,000 [3][23]. A fall toward those averages would not by itself signal an excursion: in the global models an excursion goes with a dipole of about half today’s, 4 ×10²² A m² [11 §4.4], and the Laschamp’s lowest point in GGF100k was 2.44 [24 §4.2], while the models built for the excursion put its low at about 2–5% of today’s dipole (our review of the record) [11 §4.2].

Our checks on the record: When Earth’s magnetic field falters · Is there a 6,000/12,000-year cycle? · When is the next magnetic excursion due? · Magnetic excursions, floods and extinctions, on one timeline · “Every 6,000 years?”, stop 3 of our video check

2 · Today

Weakening about 5% a century, and not speeding up

Earth’s axial dipole is about 9% weaker than in 1840, after a mean loss of 16 nT a year from 1840 to 2010 [1 pp. 1–2]. In IGRF-14, the international reference model, the full dipole moment falls from 8.32 ×10²² A m² in 1900 to 7.69 in 2025, with 7.67 forecast for 2030 (our analysis) [2][7 p. 17].

How fast the dipole weakened, five years at a time

  • Measured (IGRF-14, 1900–2025)
  • Forecast, 2025–30
  • The 1900–2025 average
0%2%4%6%8%10%1900–1905: the dipole moment weakened 4.96% a century; axial dipole 15.8 nT a year1905–1910: the dipole moment weakened 6.85% a century; axial dipole 22.0 nT a year1910–1915: the dipole moment weakened 8.93% a century; axial dipole 28.4 nT a year1915–1920: the dipole moment weakened 9.65% a century; axial dipole 30.4 nT a year9.71920–1925: the dipole moment weakened 8.61% a century; axial dipole 26.8 nT a year1925–1930: the dipole moment weakened 7.63% a century; axial dipole 24.2 nT a year1930–1935: the dipole moment weakened 5.61% a century; axial dipole 18.0 nT a year1935–1940: the dipole moment weakened 3.77% a century; axial dipole 12.2 nT a year1940–1945: the dipole moment weakened 3.92% a century; axial dipole 12.0 nT a year1945–1950: the dipole moment weakened 2.61% a century; axial dipole 8.0 nT a year2.61950–1955: the dipole moment weakened 3.49% a century; axial dipole 10.8 nT a year1955–1960: the dipole moment weakened 5.53% a century; axial dipole 15.8 nT a year1960–1965: the dipole moment weakened 5.90% a century; axial dipole 17.4 nT a year1965–1970: the dipole moment weakened 7.91% a century; axial dipole 22.8 nT a year1970–1975: the dipole moment weakened 8.62% a century; axial dipole 24.0 nT a year1975–1980: the dipole moment weakened 7.99% a century; axial dipole 21.6 nT a year1980–1985: the dipole moment weakened 9.09% a century; axial dipole 23.8 nT a year9.11985–1990: the dipole moment weakened 7.66% a century; axial dipole 19.6 nT a year1990–1995: the dipole moment weakened 6.80% a century; axial dipole 16.6 nT a year1995–2000: the dipole moment weakened 6.32% a century; axial dipole 14.5 nT a year2000–2005: the dipole moment weakened 5.68% a century; axial dipole 12.9 nT a year2005–2010: the dipole moment weakened 5.59% a century; axial dipole 11.6 nT a year2010–2015: the dipole moment weakened 5.53% a century; axial dipole 11.0 nT a year2015–2020: the dipole moment weakened 4.19% a century; axial dipole 7.6 nT a year4.22020–2025: the dipole moment weakened 4.79% a century; axial dipole 10.7 nT a year2025–2030: the dipole moment weakened 5.43% a century (IGRF-14 forecast); axial dipole 12.6 nT a year5.41900–2025 average: 6.1% a century190019251950197520002025

Weakening of the dipole moment over each five-year span, % of its strength at the start, per century

Our analysis of IGRF-14’s Gauss coefficients: the dipole moment from g10, g11 and h11 at each five-year epoch, the last bar from IGRF-14’s secular variation for 2025–30 [2]. Hover or tap a bar for its value; every value is in the table below.

The pace swings from decade to decade and is not rising: it was fastest around 1915–20 and in the early 1980s, and it is now about half the 1980s pace.

Slow here means slow for people: about 5% a century, with half today’s dipole no sooner than about 2280 on our review of the forecasts. By the core’s standards the decline is fast: “about an order of magnitude faster than the rate expected for simple decay by diffusion” [26 Introduction], a decrease that Nilsson and colleagues call rapid [27 Abstract].

Dipole moment, a century

Our analysis

−6.1%

The average pace over 1900–2025; −6.2% over 1900–50, −6.8% over 1950–2000, −5.1% over 2000–25, and −5.4% forecast for 2025–30 [2].

Tilt of the dipole axis

Our analysis

9.2°

In 2025, down from 11.4° in 1900, and 9.0° forecast for 2030: the axis is moving toward the rotation axis, not away [2]. The World Magnetic Model gives the same 9.21° [28 p. 20].

Pole travel, km, 1900–2025

Our analysis

250 vs 2,500

How far the dipole’s north pole moved, against the north dip pole, where a dip needle points straight down, from the IGRF-14 positions [7 Table 3].

Our analysis

Is it speeding up? No

Across 1900–2025 the five-year paces drift slightly down, not up. A curve fitted to the whole series loses 0.52 ×10²² A m² a century at its start and 0.50 at its end, about 6% a century throughout; fitted to 1950–2025 or 1975–2025 it shows the decline slowing [2]. That matches what Finlay and colleagues found from observatory records: “the field was decaying twice as fast in 1980 as it is today” [1 p. 2]. The 2025–30 bar is IGRF-14’s forecast, not a measurement. Our code reproduces NOAA’s official World Magnetic Model test values to 0.05 nT [29], and the script is in section 9.

The dipole every five years, and each span’s pace
EpochDipole moment, ×10²² A m²Tilt of the axisGeomagnetic north poleSpanPace, % a centuryAxial dipole, nT a year
19008.32211.39°78.69°N 68.79°W1900–19054.9615.8
19058.30111.39°78.68°N 68.75°W1905–19106.8522.0
19108.27211.41°78.66°N 68.72°W1910–19158.9328.4
19158.23611.43°78.64°N 68.57°W1915–19209.6530.4
19208.19611.44°78.63°N 68.38°W1920–19258.6126.8
19258.16011.45°78.63°N 68.27°W1925–19307.6324.2
19308.12911.47°78.60°N 68.26°W1930–19355.6118.0
19358.10711.51°78.57°N 68.36°W1935–19403.7712.2
19408.09111.53°78.54°N 68.51°W1940–19453.9212.0
19458.07511.53°78.54°N 68.53°W1945–19502.618.0
19508.06511.53°78.54°N 68.85°W1950–19553.4910.8
19558.05111.54°78.54°N 69.16°W1955–19605.5315.8
19608.02811.49°78.58°N 69.47°W1960–19655.9017.4
19658.00511.47°78.61°N 69.85°W1965–19707.9122.8
19707.97311.41°78.67°N 70.18°W1970–19758.6224.0
19757.93911.31°78.76°N 70.47°W1975–19807.9921.6
19807.90711.19°78.88°N 70.76°W1980–19859.0923.8
19857.87111.03°79.05°N 70.90°W1985–19907.6619.6
19907.84110.86°79.21°N 71.13°W1990–19956.8016.6
19957.81410.68°79.39°N 71.42°W1995–20006.3214.5
20007.79010.46°79.61°N 71.57°W2000–20055.6812.9
20057.76810.25°79.82°N 71.81°W2005–20105.5911.6
20107.7469.98°80.08°N 72.21°W2010–20155.5311.0
20157.7249.69°80.38°N 72.61°W2015–20204.197.6
20207.7089.41°80.65°N 72.68°W2020–20254.7910.7
20257.6909.21°80.85°N 72.76°W2025–20305.4312.6
2030 (forecast)7.6699.01°81.05°N 72.96°W———

Our analysis of IGRF-14 [2]. Rates per century over longer spans, as a share of the starting value: 1900–1950 −6.17% (−0.513 ×10²² A m²); 1950–2000 −6.82% (−0.550 ×10²² A m²); 2000–2025 −5.13% (−0.400 ×10²² A m²); 1900–2025 −6.07% (−0.505 ×10²² A m²); 2025–2030 −5.43% (−0.418 ×10²² A m²). The pole’s latitude is geodetic, as in the IGRF-14 paper’s Table 3, and agrees with that table to 0.01° [7 Table 3].

The 2025 World Magnetic Model is working as forecast

A year in, its errors “increased by less than 1% over the past year and remained well below the limits specified for the WMM”, and the north and south dip poles moved at 36 and 9 km a year [30 p. 3]. NOAA: their positions and speeds proved “very close to those predicted”, and WMM2025 and its high-resolution twin “prove to be accurate models of Earth’s magnetic field” [31]. Our video check’s stop 9 rates “The 2025 model already failing?” False.

What the models’ makers expect

“The dipole field is forecast to continue to fall in strength over the next five years and weaken relative to the non-dipolar contributions” [7 p. 17]. Finlay and colleagues expect the decay “to continue, at least for the next few decades” [1 p. 1]. ESA: “Over the last 200 years, the magnetic field has lost around 9% of its strength on a global average” [32].

Our checks on today’s field: When Earth’s magnetic field falters, section 4 · When is the next magnetic excursion due?, section 3 · S0 News, October 20, 2026 · “15% gone, ten times faster?”, stop 6 · “The 2025 model already failing?”, stop 9 · “55% of the field gone?”

3 · The South Atlantic Anomaly

Deeper, drifting west and splitting in two, as such weak spots have before

Over South America and the South Atlantic the field is weakest: “In the SAA the intensity of the field is about one-third of that near the magnetic poles” [30 p. 13]. In 2025 its minimum was 22,071 nT, in northeastern Argentina, and it “has deepened by around 150 nT over the last five years and moved westward at around 20 km/year” [7 p. 13].

The anomaly in IGRF-14, 1900–2030

  • The minimum; the area below 26,000 nT
  • The area below 32,000 nT
  • Forecast, 2025–30
The lowest field at Earth’s surface, nT22,00023,00024,00025,00019001950200020301900: 25,422 nT1925: 25,176 nT1950: 24,504 nT1975: 23,641 nT2000: 22,874 nT2025: 22,071 nT2030 (forecast): 21,907 nT25,422 nT in 190022,071 nT in 2025Share of Earth’s surface below each level0%5%10%15%20%25%19001950200020301900: 9.0% of Earth’s surface1925: 9.9% of Earth’s surface1950: 12.3% of Earth’s surface1975: 16.1% of Earth’s surface2000: 19.0% of Earth’s surface2025: 21.4% of Earth’s surface2030 (forecast): 22.0% of Earth’s surface1900: 0.8% of Earth’s surface1925: 1.0% of Earth’s surface1950: 1.8% of Earth’s surface1975: 3.1% of Earth’s surface2000: 5.5% of Earth’s surface2025: 7.7% of Earth’s surface2030 (forecast): 8.2% of Earth’s surfaceBelow 32,000 nT: 21.4% in 2025Below 26,000 nT: 7.7%
Our analysis of IGRF-14 at the WGS84 surface: the minimum refined to 0.05°, areas from a 1° grid with exact cell areas; every area is one region joined to the minimum [2]. The IGRF-14 paper gives the 2025 minimum as 22,071 nT at 26.22°S, 60.03°W [7 p. 13]. Hover or tap a dot for its value.
The anomaly at nine dates, with its area below three levels
YearMinimum, nTWhere% of Earth below 32,000 nTBelow 26,000 nTBelow 25,000 nT
190025,42222.0°S 37.0°W9.020.760.00
192525,17624.0°S 42.3°W9.881.010.00
195024,50425.9°S 45.5°W12.331.830.59
197523,64126.2°S 51.0°W16.153.091.68
200022,87426.2°S 54.6°W19.035.503.06
201422,41326.3°S 57.5°W20.256.794.72
202022,22426.2°S 59.0°W20.827.265.31
202522,07126.2°S 60.0°W21.407.725.83
2030 (forecast)21,90726.2°S 61.0°W22.008.166.30

Our analysis of IGRF-14 [2]. A separate low south-west of Africa appears in 2006, under 1 nT deep below the saddle that joins it to the main minimum; it is 28 nT deep in 2010, 84 in 2015, 177 in 2020 and 285 in 2025, when it reaches 23,462 nT at 40.75°S 0.75°W. IGRF-14 runs to degree 10 before 2000 and degree 13 from 2000.

The minimum, 1900–2025

Our analysis

−13%

From 25,422 nT off Brazil in 1900 to 22,071 nT in 2025, while the low point moved about 2,400 km west and south, about 19 km a year [2].

Growth since 2014

Our analysis

+0.9% of Earth

The area below 26,000 nT, now 7.7% of the surface in IGRF-14 [2]; the Swarm team found the same 0.9% [33 Abstract], “an area nearly half the size of continental Europe” [34].

A second low, 2025

Our analysis

285 nT

How far the second low, south-west of Africa, sits below the saddle joining it to the main anomaly in IGRF-14, up from 28 nT in 2010 [2]. ESA, in 2020: “Over the past five years, a second centre of minimum intensity has emerged southwest of Africa” [32]; in CHAOS-7 it “is seen even if the field is truncated at spherical harmonic degree 9” [35 p. 26].

What it affects: satellites, not life on the ground

“satellites passing over the region are faced with higher doses of incoming radiation. This can lead to malfunctions or damage to critical hardware, and even blackouts.”

European Space Agency (2025) [34]

“No astronomical or calibration observations are possible (except for some specialized and infrequently executed WFC3 observations) during spacecraft passage through the SAA”

Space Telescope Science Institute, on Hubble [36]

“The International Space Station, which is in low-Earth orbit, also passes through the SAA. It is well protected, and astronauts are safe from harm while inside.”

Swarm, the satellites that map it, feel it too: “the highest concentration of SEUs”, single-event upsets in their electronics, “is clearly observed at mid and low latitudes in the South Atlantic Anomaly weak field region” [35 p. 24]. At the surface, NASA says, “the SAA creates no visible impacts on daily life” [37].

What its history suggests

LLSVP is the large low-shear-velocity province, a vast region of unusual rock at the base of the mantle under Africa.

“We propose that the recurrence of low field strengths reflects core flux expulsion promoted by the unusual core–mantle boundary (CMB) composition and structure beneath southern Africa defined by the African large low shear velocity province (LLSVP).”

Tarduno et al. (2015), from records of AD 1000–1600 [38 Abstract]

“the SAA is not a single occurrence but rather, the latest in a series of recurring weaknesses in the field in this region”

Engbers et al. (2022), from lavas 10 to 8 million years old on Saint Helena [39 Abstract]

“The analogy implies that the South Atlantic Anomaly will likely disappear in next few hundred years, accompanied by a return to a more symmetric field configuration and possibly, a strengthening of the axial dipole field.”

Our analysis

What it does and doesn’t portend: debated

Weak spots like this one have come and gone in the South Atlantic for millions of years: directions on Saint Helena “did indeed vary anomalously between ∼8 million and 11 million years ago” [40], and Tarduno and colleagues suggest the region “may have been a steady site for flux expulsion, and triggering of geomagnetic reversals, for millions of years” [38 Abstract]. Whether today’s anomaly heralds a transition is argued both ways. Pavón-Carrasco and De Santis read reversed patches beneath it as typical of “the preparation phase of a geomagnetic transition”, and add that today’s dipole “is not so low” next to the Holocene or the whole Brunhes, “although the rate of decay is similar to that given by previous documented geomagnetic reversals or excursions” [41 Abstract]; but a 185-year trend and a rate averaged over millennia are not the same quantity (our check of a 2026 paper). ESA calls the dip “well within what is considered normal levels of fluctuations” [32]. Nilsson and colleagues, whose ancient analogue has the anomaly fading, note that the dipole “is still decreasing without showing signs of slowing down, which is not consistent with the proposed ancient analog” [27 Discussion]; it is still falling, though its pace has eased since about 1980 [1 p. 2]. None of them dates an excursion.

Our checks on the anomaly: The core field accelerates most near the equator (“SAA, West Pacific, the Triangles”, Misleading) · S0 News, October 20, 2026 (“America loses protection fastest”, Misleading) · When Earth’s magnetic field falters, section 4

4 · Timescales

Centuries to millennia, not two or three generations

The field’s big changes run far longer than a human life. In the last four reversals the field’s direction took about 7,000 years on average to change [42 Abstract], and the whole last reversal “spans at least 22 ka and perhaps as long as 30 ka”, ending “rapidly—in less than 4000 years” [4 Discussion].

How long things take, on one ladder

  • The field’s changes
  • Human timescales
  • The episode’s claim
101001,00010,000100,0001 millionA human generationA human generation: 27 years years27 years“Two or three generations” (the episode’s claim)“Two or three generations” (the episode’s claim): 50–90 years50–90Direct measurements of the field’s strength, since 1840Direct measurements of the field’s strength, since 1840: 185 years185Halving today’s dipole, on any of five extrapolationsHalving today’s dipole, on any of five extrapolations: 255–1,355 years255–1,355Laschamp excursion: its directional changesLaschamp excursion: its directional changes: 640 years640Laschamp: its dipole’s fallLaschamp: its dipole’s fall: 1,500–1,900 years1,500–1,900Laschamp: its expression at Earth’s surfaceLaschamp: its expression at Earth’s surface: about 2,000 yearsabout 2,000Laschamp: its underlying process, in the coreLaschamp: its underlying process, in the core: about 5,000 yearsabout 5,000A reversal’s change of direction, average of fourA reversal’s change of direction, average of four: 7,000 years7,000The whole Matuyama–Brunhes reversalThe whole Matuyama–Brunhes reversal: 22,000–30,000 years22,000–30,000Average time between reversals, last 25 million yearsAverage time between reversals, last 25 million years: about 200,000 yearsabout 200,000Time since the last reversalTime since the last reversal: 773,000 years773,000

Years, on a logarithmic scale: each step is ten times the last

Our analysis: each rung from the source in the table below. Bars are ranges, dots single values; short ranges are drawn at a minimum length.

The 7,000 years is how long the direction took to change [42 Abstract], within a process of 22,000–30,000 years [4 Discussion], as stop 10 of our video check now says.

The fastest changes ever claimed, and where they stand

Where and whenWhat was claimedWhere it stands
Steens Mountain, Oregon, a reversal in lava flows“extremely rapid field directional changes (6 degrees per day)” inside a single lava flowIts authors’ later sampling “weigh[s] against the hypothesis that the geomagnetic field cause the unusual intraflow fluctuations”, though the alternative, reheating by later flows, “remains also ad hoc” [43 Abstract]
Sheep Creek Range, Nevada, 15.6 million years agoA 53° turn “at an average rate of approximately 1°/week”, read from a reheated lava flowIt “reinforces the controversial hypothesis that geomagnetic change during a polarity reversal can be much faster than normal” [44 Abstract]
Sulmona Basin, Italy, the last reversalA flip “in less than a century” (2014), then one that “lasted less than 13 ± 6 yr” unless the record was overprinted (2016) [45]Disputed: “this section does not carry a reliable high-resolution record of the geomagnetic field” [46 Abstract]; the original authors replied in 2019 [47]
Computer models of the geodynamoDirectional changes up to about 10° a year, “almost 100 times faster than current changes”Simulations; reported rates in rocks of up to 1° a year are “controversial” [48 Abstract]

All four concern the field’s direction at one place. The dipole’s strength still takes centuries to millennia to collapse.

Our analysis

Two or three generations is off the scale

S0 News said an excursion brings “almost that entire same amount of change hitting only two or three generations” [10 00:03:02–00:03:07], 50 to 90 years. On the ladder that sits an order of magnitude below the Laschamp’s directional changes, 640 years [26 Results and discussion], and its dipole took about 1,500–1,900 years to fall at the rates a 2026 study fitted (our arithmetic) [25 Table 1]. Our check of the paper behind that episode found that a fall as deep within 50–90 years would need 13 to 23 times the steepest rate in it, and rated “Hits two or three generations” False. The extrapolations on our review of the forecasts put even a halving of the dipole at least 255 years away.

The ladder’s rungs and their sources
RungYearsSource
A human generation27 years[49 Abstract]
“Two or three generations” (the episode’s claim)50–90[10 00:03:02–00:03:07]
Direct measurements of the field’s strength, since 1840185[1 p. 2]
Halving today’s dipole, on any of five extrapolations255–1,355Our analysis on our review of the forecasts, from IGRF-14 [2] and the 1840–2010 mean rate [1 p. 2]: AD 2280–3380
Laschamp excursion: its directional changes640[26 Results and discussion]
Laschamp: its dipole’s fall1,500–1,900[25 Table 1]
Laschamp: its expression at Earth’s surfaceabout 2,000[11 Abstract]
Laschamp: its underlying process, in the coreabout 5,000[11 Abstract]
A reversal’s change of direction, average of four7,000[42 Abstract]
The whole Matuyama–Brunhes reversal22,000–30,000[4 Discussion]
Average time between reversals, last 25 million yearsabout 200,000[22 §5]
Time since the last reversal773,000[4 Teaser]

A generation of 25–30 years makes “two or three generations” 50–90 years. The Laschamp’s fall is our arithmetic from the decay rates a 2026 study fitted, on our check of that paper. The average time between reversals is 1 ÷ 5 per million years.

Our checks on timescales: Excursions weaken the field faster than reversals do, but over millennia, not generations · S0 News, October 23, 2026 · When is the next magnetic excursion due? · “A magnetic pole flip in the 2040s?”, stop 10 · “A 2025 model of pole flips”

5 · Life under a weak field

No extinctions on the record, and a modest change in radiation at the ground

During a reversal “the magnetic field at the surface of the Earth decreases to about 10% of its current value”, yet a review finds “no clear picture of a geomagnetic link” to the evolution of life [12 Abstract]. At the ground the air, not the field, does most of the shielding.

Cosmic rays at the ground, today

0.3 of 3.0 mSv

The worldwide average yearly dose from cosmic rays [15 ¶290], out of 3.0 mSv from all natural sources [15 ¶284].

With no field at all

Our analysis

+0.1 mSv a year

Our upper bound for the extra dose at sea level at low latitudes, outdoors: about a third to a half more than today [14 ¶20–25][50 Results].

At flight altitude, at the Laschamp’s low

4.5 mSv

From a solar particle storm like February 1956’s, in about four hours at 12 km, at the worst-placed point: about 1.5 times aircrew’s average yearly dose [51 §5].

Our analysis

How we bounded the dose at the ground

With no field, every place would get today’s polar-cap dose. At sea level the directly ionizing part, mostly muons, varies only about 10% with latitude, while the neutron part is four to five times lower at the equator than near the poles, where it is about 11 nSv an hour [14 ¶20–25]; Sato’s model has the absorbed dose rate in outdoor air at sea level, the measure behind UNSCEAR’s ionizing values, rising from 26.8 to 31.8 nSv an hour as the cutoff falls; it runs about 9% above the effective dose [50 Results]. So a low-latitude dose of about 29–33 nSv an hour would rise to about 43–44: roughly 0.1 mSv a year more, outdoors, 3–4% of the natural total. Altitude matters more: at 4 km the muon dose alone is “approximately double that at sea level” [15 ¶188]. In a reversal the surface field keeps about a tenth of its strength [12 Abstract]; at the Laschamp’s low, cutoffs still reached 4 GV and some routes were better shielded than today [51 §6], so this is an upper bound.

On flights and on the ground, in the modellers’ words

Larsen and colleagues ran cosmic-ray paths and doses through the Laschamp excursion of 41,000 years ago.

“While we are well protected on the ground by the thick atmosphere, the level of CR-induced radiation can be much higher at aviation altitudes”

“low-latitude flights, normally well shielded under modern conditions, can experience significant increases in dose rates; in contrast, some high-latitude flight routes may observe decreases in radiation exposure”

“Such geomagnetic excursions are not likely in the near future”; still, “The effects of a weakening and reversal of the magnetic field on the atmosphere and environment can be dramatic, yet still largely unknown”

Ilya Usoskin and Pauli Väisänen; the second sentence, Usoskin; University of Oulu (2026) [52]

The isotopes record the weak spells

Cosmic rays make beryllium-10 and carbon-14 in the air, so their production climbs when the field is weak. “The largest ¹⁰Be overproduction episodes are related to dipole field collapses (below a threshold of 2 × 10²² Am²)”, at the last reversal, the Laschamp and the Iceland Basin event [53 Abstract]; during the Laschamp “cosmogenic isotope production increased dramatically” [54 Abstract]. These are records of weak fields, not of extinctions.

The extinction question

Asked “Is there any danger to life?”, the British Geological Survey answers “Almost certainly not” [13]. The argued case, a 2021 paper tying the Laschamp to climate shifts and extinctions [55], drew formal comments; one, by ten researchers, holds that the proposition “is unproven from the current archaeological, paleoanthropological, and genetic records” [56]. Our review of the October 22 episode rates “Excursions take out much of life” False.

A weaker field does let more particles in, but two things blamed on it in the episodes don’t follow. The 2026 ozone hole tracks the stratosphere’s temperature, not geomagnetic activity (“Weak field explains ozone hole”, False), and the October 20 episode’s laboratory “hypomagnetic” studies used fields about 50 to 350 times weaker than those their subjects live in (“Earth nears lab-like low fields”, False).

Our checks on life and the field: S0 News, October 22, 2026 · Magnetic excursions, floods and extinctions · Excursions weaken the field faster than reversals do · S0 News, October 24, 2026 · Storm ozone loss, yes. Pole shift, no. · The 2026 Antarctic ozone hole · Six volunteers in a near-zero field · Mice in a near-zero field · Mole-rats notice a missing magnetic field · Fruit flies without Earth’s field · Bats in total darkness · “Ozone loss as the ‘extinction’ mechanism”, note 14

6 · What physics rules out

No 90° turn of the crust, and no solar micronova

Two ideas in the video sit outside physics as measured: a crust that turns 90° in a day (stop 17), and a Sun that blasts off its outer layers in a “micronova” (stop 15). Both are rated False.

A 90° slide in a day

Our analysis

7 × 10²⁶ J

Kinetic energy to turn a 100-km-thick lithosphere 90° in 24 hours, from our review of Earth’s energy stores; each megapascal of drag at its base would cost about 4 × 10²⁷ J more.

All of Earth’s heat loss, in a day

4 × 10¹⁸ J

Earth loses heat at “47±2 TW” [16 Abstract]; over 24 hours that is about 4 × 10¹⁸ J, less than a hundred-millionth of the slide’s need (our arithmetic).

How fast the pole can wander

≤3° per million years

“Paleomagnetic data constrain TPW to ≤3° million year−1 (Ma−1) during the Phanerozoic (ca. 539 Ma ago to present)” [17 Introduction]; the fastest well-dated episode, 832–821 million years ago, ran at more than 5° per million years [17 Abstract].

Micronovae happen on white dwarfs

They are real and were announced in 2022: “Both types of explosions occur on white dwarfs, dead stars with a mass about that of our Sun, but as small as Earth”, micronovae on “some white dwarfs with strong magnetic fields, which funnel material towards the star’s magnetic poles” [57]. The bursts, about a millionth of a nova’s energy, come from “thermonuclear runaway events in magnetically confined accretion columns” [58 Abstract]. The Sun is not a white dwarf and has no companion feeding it (our check).

Magnetic poles are not geographic poles

The north dip pole wanders tens of kilometres a year and the dipole axis tilts by fractions of a degree a decade; neither moves the ground. The geographic poles move only by true polar wander, a few degrees per million years at most [17 Introduction]. “Pole shift” in the episodes means the magnetic poles; the video’s 90° turn would move the crust itself.

Our checks on the physics: Earth’s energy stores are big enough, but none could slide the crust 90° in a day · ECDO’s rapid Earth flip runs against measurements · Polar forests, yes. Pole shifts, no. · “A solar micronova?”, stop 15 · “Isotopes from the Sun?”, stop 16 · “Einstein and the 90° flip”, stop 17

7 · The real near-term risk

Solar storms are the hazard to plan for, and the field’s slow decline barely changes them

The risk from space that matters within a lifetime is space weather: eruptions on the Sun that shake Earth’s magnetic field, flood near-Earth space with particles or black out radio. A storm as large as 1859’s is rare, but, as Lloyd’s put it, “it is almost inevitable that one will occur eventually” [59 p. 4]. Grids, satellite operators and airlines plan for it, with warnings of days from the Sun and minutes from monitors in space.

How storms are graded

NOAA’s three scales “list possible effects at each level” and “show how often such events happen” [20]. Our count of storm days in GFZ’s Kp record since 1932 is close to NOAA’s for G1 to G3 and lower for G4 and G5 [60].

NOAA’s three scales, with our count of storm days

LevelMeasured bySome of NOAA’s effectsHow often, per 11-year cycle (NOAA)Our count: days per cycle, GFZ’s Kp
Geomagnetic storms, by Kp, the planetary index of magnetic disturbance in each three hours
G5 ExtremeKp 9Some grids “may experience complete collapse or blackouts”; transformers may be damaged; aurora to about 40° geomagnetic latitude4 (4 days)3.2
G4 SevereKp 8, including 9−Protective systems may trip out key parts of the grid; aurora to about 45°100 (60 days)44.5
G3 StrongKp 7Voltage corrections may be needed; drag on low-orbit satellites may increase; aurora to about 50°200 (130 days)127
G2 ModerateKp 6Voltage alarms on high-latitude grids; long storms may damage transformers; aurora to about 55°600 (360 days)356
G1 MinorKp 5Weak fluctuations on power grids; aurora at high latitudes1,700 (900 days)874
Solar radiation storms, by the flux of particles above 10 MeV, in particle flux units (pfu)
S5 Extreme100,000 pfuSatellites “may be rendered useless”; an unavoidable radiation hazard to astronauts on spacewalks; radiation risk on high-latitude flightsFewer than 1—
S4 Severe10,000 pfuSatellite memory and imaging problems; HF radio blacked out over the polar regions3—
S3 Strong1,000 pfuSingle-event upsets in satellites likely; radiation risk on high-latitude flights10—
S2 Moderate100 pfuInfrequent single-event upsets; elevated radiation risk on high-latitude flights25—
S1 Minor10 pfuMinor effects on HF radio in the polar regions50—
Radio blackouts, by the peak X-ray flux of a solar flare, 0.1–0.8 nm
R5 ExtremeX20, 2 × 10−3 W m−2HF radio blacked out on the whole sunlit side of Earth for hoursFewer than 1—
R4 SevereX10HF radio blacked out on most of the sunlit side for one to two hours8 (8 days)—
R3 StrongX1A wide-area HF blackout, radio contact lost for about an hour175 (140 days)—
R2 ModerateM5A limited HF blackout, contact lost for tens of minutes350 (300 days)—
R1 MinorM1, 10−5 W m−2Minor degradation of HF radio, occasional loss of contact2,000 (950 days)—

Levels, measures, effects and frequencies from NOAA’s scales, which count events, with storm days in brackets [20]. Our count: days whose highest Kp reached the level or more, averaged over the complete solar cycles 17–24 (late 1933 to late 2019), from GFZ’s definitive Kp; cycle 24 had no G5 day [60]. The script is in section 9.

The storms that did damage

Seven storms, from 1859 to 2024

StormSizeWhat it did
1859 September, the Carrington EventAbout −900 nT, reconstructed [61 §2.2.1]Telegraph systems were disrupted worldwide [62]; aurora was seen across the Caribbean and Mexico and in Chile [63 §3.3], and in southern Japan [64 §3.1].
1921 May, the New York Railroad Storm−907 ± 132 nT, reconstructed [65]Fires destroyed the railroad station at Brewster, New York [66 p. 525][67 p. 3], and hit the telegraph station at Karlstad, Sweden [68 p. 356].
1989 March−589 nT; Kp 9Hydro-Québec’s grid collapsed in about 90 seconds, and about six million people lost power for about nine hours [69][70]. In New Jersey, the generator step-up transformer of the Salem No. 1 nuclear plant was found with “Charred winding series connections” [71 pp. 38–39].
2003 October, the Halloween storms−353 and −383 nT; Kp 9An outage of under an hour in Malmö, Sweden [72 p. 1782][73], and damage that led several large transformers in South Africa to fail in the following months [74][75]. NASA Goddard’s mission operations team found that “approximately 59% of the Earth and Space science missions were impacted”, NOAA reported, and the storms are “suspected to have caused the loss of the $640 million ADEOS-2 spacecraft” [76 p. 1]; the FAA “issued their first ever advisory” on radiation at flight altitudes [76 p. 18].
2012 July, a near missNone on Earth; about −500 nT modelled, up to −1,182 nTA coronal mass ejection launched at about 2,500 km/s, aimed away from Earth, reached the STEREO-A spacecraft about 19 hours later. Aimed at Earth, it “would certainly have produced a geomagnetic storm that was comparable to the largest events of the twentieth century” [77 Abstract].
2022 February, the Starlink loss−66 nT; Kp 5+, a G1 stormSpaceX lost 38 of the 49 satellites it had just launched: the minor storm raised the drag on them “up to 50 percent higher than during previous launches”, near their deployment perigee of about 210 km [78][79 Abstract].
2024 May, the Gannon storm−406 nT, provisional; Kp 9; daily Ap 271, the second-highest since 1932The first day of Kp 9 since October 2003. NERC found that North America’s bulk power system “remained stable throughout the event”, its operators alerted “six hours before the onset” [80 pp. 1–2]. “No transformer outages or degradations from the Gannon Storm were reported into TADS”, NERC’s outage database, though on the same page “The failure of a surge arrester caused the outage of a transformer” [80 p. 15].

Dst, Kyoto’s storm index from four low-latitude observatories: final values for 1989, provisional for 2022 and 2024 [81]. Kp and daily Ap from GFZ, with our count and ranking of every day since 1932 [60]. The 1859, 1921 and October 2003 values and the 1859–2003 effects are as on our review of the record, section 5, with its sources; our check of the Sun’s two worst cases covers 1859.

How often a Carrington-class storm may come

The chance of a Carrington-class storm within ten years, nine estimates

  • As published
  • Our conversion of a published rate
  • Its uncertainty range, where given
0%10%20%30%40%Riley 2012power law; Dst below −850 nTRiley 2012, power law; Dst below −850 nT: about 12%; stated; no rangeabout 12%Love 2012counting storms since 1859; −Dst of 1,760 nT or moreLove 2012, counting storms since 1859; −Dst of 1,760 nT or more: 6.3% (under 0.1–23); 95% credibility, under 0.1% to 23%6.3% (under 0.1–23)Kataoka 201389-year storm list from KakiokaKataoka 2013, 89-year storm list from Kakioka: 4–6%; range given4–6%Love and others 2015lognormal; −Dst of 850 nT or moreLove and others 2015, lognormal; −Dst of 850 nT or more: 10.7% (4.1–21.4); our conversion of 1.13 a century, 95% range 0.42–2.4110.7% (4.1–21.4)Riley and Love 2017power lawRiley and Love 2017, power law: 10.3% (0.9–18.7); 95% confidence10.3% (0.9–18.7)Riley and Love 2017lognormalRiley and Love 2017, lognormal: 3.0% (0.6–9.0); 95% confidence3.0% (0.6–9.0)Moriña and others 2019Weibull waiting times; Dst below −850 nTMoriña and others 2019, Weibull waiting times; Dst below −850 nT: 0.92% (0.46–1.88); 95% confidence0.92% (0.46–1.88)Chapman and others 2020aa index since 1868Chapman and others 2020, aa index since 1868: 6.8%; our conversion of 0.7% a year; no range6.8%Lloyd’s and AER 2013historical aurora recordsLloyd’s and AER 2013, historical aurora records: 6.4% (3.9–9.5); our conversion of a 150-year return period, 100–2506.4% (3.9–9.5)

Chance within a decade

Estimates as published, or converted by us to a chance per decade from a published rate or return period; the ranges are 95% intervals except Kataoka’s and Lloyd’s, which are stated ranges. Most are from abstracts; Moriña’s and Chapman’s papers and Lloyd’s report were read in full. Sources: Riley 2012 [82 Abstract]; Love 2012 [83]; Kataoka 2013 [84 Abstract]; Love and others 2015 [85 Abstract]; Riley and Love 2017 [86 Abstract]; Moriña and others 2019 [87 Abstract]; Chapman and others 2020 [88 Plain Language Summary]; Lloyd’s and AER 2013 [59 p. 8].

The spread is the finding

Riley and Love put the chance of a Carrington-class storm within a decade at 10.3% with one statistical model and 3.0% with another, and cautioned that “our results depend crucially on (1) how we define an extreme event, (2) the statistical model used to describe how the events are distributed in intensity”, the fitting technique and the data [86 Abstract]. For planning, NERC builds its standard on a 1-in-100-year storm and judges that “the Carrington event is likely less frequent than 1-in-100 years” [89 p. 8]; the UK plans for a Carrington-scale storm lasting one to two weeks [90 p. 137].

How bad it could be, by the kind of study

Estimates of the damage depend on the kind of study that makes them, so they are listed here by kind, not averaged. A review of the economics found “disagreement as to the severity of the technological footprint” [91 Abstract].

Five kinds of answer

KindStudyWhat it found
Worst-case scenarioNational Research Council workshop report, 2008An estimate “of $1 trillion to $2 trillion during the first year alone” for the costs of a “severe geomagnetic storm scenario”, with recovery times of 4 to 10 years [92 p. 4]
Insurance scenarioLloyd’s and AER, 2013For a Carrington-level storm, 20–40 million people in the United States at risk of long outages, “with durations of 16 days to 1-2 years”; new transformers a lead time of “a minimum of five months”; a cost of “$0.6-2.6 trillion” [59 p. 4]
Engineering assessmentNERC, 2012“The most likely consequence” is “the loss of voltage stability”: restoration “a matter of hours to days”, while replacing transformers takes “a number of months”, and “the failure of a large numbers of transformers would have considerable impacts on portions of the system” [71 p. iv]; above 500 kV, most transformers “typically take one to three years to acquire” [71 p. 111]
Planning scenario: the UK government’s reasonable worst caseUK National Risk Register, 2025For a Carrington-scale storm: urban outages “recovered in a matter of hours”; where transformers in remote coastal areas need replacing, “several months”; satellite navigation back in “several days, with a small number of satellites non-recoverable” [90 p. 138]
Measured outcomeNERC staff review of May 2024, published 2026The grid stayed stable [80 p. 1]; the highest induced current in the United States was 175.7 amps [80 p. 6]; five generators were affected [80 p. 14]; by NERC’s severity index the day ranked the “68th most impactful day of 2024” [80 p. 12]

Warning and protection

Days from the Sun, minutes from L1

The fastest coronal mass ejections “can reach our planet in as little as 15-18 hours” [93]; a 2008 workshop report found that NOAA’s forecasters could, with moderate confidence, “predict the occurrence probability of a geomagnetic storm or an X-class flare 1 to 3 days in advance” [92 p. 4], and in May 2024 grid operators had six hours’ notice [80 p. 2]. The last word comes from monitors a million miles sunward, at L1, which give “15 to 60 minutes advanced warning” [93]. NOAA’s SOLAR-1 has been the primary monitor there since June 10, 2026 [94]. DSCOVR was taken offline in July 2025 “until further notice”, with ACE as NOAA’s primary satellite meanwhile, and we found no later notice of its return [95].

Rules for grids

Since 2015, North America’s grid operators have had to run operating procedures for geomagnetic storms [80 p. 3][96]. Under TPL-007 they must also test their systems against a benchmark storm built on “a reference peak geoelectric field amplitude of 8 V/km”, scaled for geomagnetic latitude and the ground’s conductivity [97 Attachment 1]. NERC’s 2012 assessment found that replacing failed extra-high-voltage transformers “is not a small undertaking”, but that the loss of a few “would rarely challenge bulk power system reliability” [71 p. iii].

Does a weaker field leave Earth more exposed? Barely, and storms don’t grow

S0 News: “Earth is letting in more space particle energy as our shield weakens” [5 00:01:57–00:02:02]. The direction is right; the size is not.

A century of the field’s weakening, against one storm or one solar cycle

  • The field’s change, 1900–2025 (counts at the ground: 1913–1997)
  • A storm, or a solar cycle

Size of the change, in each row’s own unit, on a logarithmic scale: each step is ten times the last

On a logarithmic scale the gap between a row’s blue and orange marks shows how many times larger the Sun’s effect is than the field’s century of change. Our analysis: production from our re-run of Kovaltsov and colleagues’ model [18]; the magnetopause and the oval from the scalings in Glassmeier and colleagues, given for the polar cap’s boundary and applied here to the oval’s edge [19 p. 3669], with IGRF-14 [2]. Counts at the ground: the change in each station’s count rate, from Herbst and colleagues’ traced cutoffs [98 Table 2]; the Oulu station’s data [99]; Tsumeb, in Namibia, over 2006–2014 [100 p. 7]. May 2024’s oval from our check of that storm. Hover or tap a mark for its value.
Our analysis

A century of weakening, in numbers

From 1900 to 2025 the dipole weakened 7.6% [2]. In simple scalings, the cutoff that cosmic rays must beat to reach the air, averaged over the globe, fell 7.6% with the dipole alone and 8.8% with the field’s shape included; global production of radiocarbon, a gauge of the cosmic rays reaching the air, rose 3.9–4.5% in a re-run of Kovaltsov and colleagues’ model [18]; the magnetopause, where the field holds off the solar wind, moved 2.6% closer; and the auroral oval’s edge moved 0.35–0.91° toward the equator in the Siscoe–Chen scaling, or 0 to 2.8° across the three scalings Glassmeier and colleagues compare, which they give for the polar cap’s boundary and we apply to the oval’s edge [19 p. 3669]. The Sun moves the same things much further. Production in the same model swings by about ×1.8 between solar minimum and maximum, close to the paper’s own “factor of two” over 1951–2010 [18 p. 5]; at the Oulu neutron monitor, cosmic-ray counts fall 10–16% from solar minimum to maximum, and fell 25% in the October 2003 storm [99]. At ground level the field’s century of change depends on the place: in Herbst and colleagues’ traced cutoffs, counts changed by about 1% or less at Kiel and Moscow over 1913–1997, but rose about 19% at Mexico City and fell about 12% in the British Virgin Islands (our arithmetic from their Table 2) [98 Table 2], as much as or more than the about 5% by which a low-latitude monitor, Tsumeb in Namibia, varied with the Sun over 2006–2014 [100 p. 7]. The dose involved stays small: even with no field at all, the dose at sea level would rise by at most about 0.1 mSv a year (section 5). In May 2024 a single storm moved the oval’s edge from the usual 55–65° to about 30–36° (our check of the May 2024 aurora). Glassmeier and colleagues concluded that the dipole’s secular change “is of minor importance compared to the overall solar wind induced variability of the magnetosphere” [19 p. 3676], and that the ring-current disturbances storms are measured by “do not increase with decreasing dipole moment” [19 Abstract]. Storms themselves range from 66 nT (February 2022) to 589 nT (March 1989) [81], and NOAA’s radiation scale spans a factor of 10⁴ [20].

Our analysis

Where you live matters more, and eastern and central North America gained

The dipole’s axis has tilted back toward the rotation axis and toward Siberia, so cities from the Midwest to the East Coast now sit 1.4–2.2° lower in geomagnetic latitude than in 1900, and their cutoffs rose 5–17%: more shielding, not less [2]. Counting the oval’s own small shift, they sit 0.8–1.5° farther from a storm-time oval edge near 50°; Helsinki and Moscow sit 0.8–1.2° closer. Studies that trace particles through the field agree on the direction there: among high-latitude cosmic-ray stations since 1950, “For the American, Canadian, and European stations, a systematic increase of cutoff rigidities is observed”, which the authors tie to an anomalous region over the northern Atlantic [101 p. 4]; farther east, in Finland and Russia, cutoffs fell, as the table shows and as Herbst and colleagues traced for Moscow [98 Table 1]. And “the auroral zones are currently drifting away from North America and towards Siberia” [21 §3.1].

Geomagnetic latitude, 1900 and 2025

City19002025ChangeCosmic-ray cutoff
Anchorage60.77°61.77°+1.00°−19%
Seattle53.25°52.85°−0.40°−4%
Denver48.34°47.14°−1.20°+1%
Chicago52.33°50.54°−1.79°+8%
New York51.85°49.73°−2.12°+11%
Quebec City57.99°55.82°−2.17°+17%
London54.12°53.17°−0.94°+1%
Helsinki57.41°57.58°+0.17°−9%
Moscow51.00°51.52°+0.51°−12%

Our analysis: latitude from the dipole of each epoch in IGRF-14, and the change in the vertical cutoff rigidity at 20 km; a higher cutoff means more shielding [2]. The apex latitude, which follows the field lines, moved more in the North-East: Ottawa’s from 58.3° to 53.5°.

What does not follow

The documentary clip in the video that warns of “maybe years to fix all the blackouts” [6 11:47] is rated at stop 1, “A ‘small’ flare?” (Partly true): years appear in the worst-case scenarios and where many large transformers would need replacing, while the engineering assessments expect most outages to last hours to days. The video goes further: “with Earth’s magnetic field weakening … it’s going to take less and less extreme of a blast from the sun to do the same level of damage here on Earth” [6 12:25–12:38]. Storms do not grow as the dipole weakens, and over eastern and central North America exposure fell. That idea, that a weaker field amplifies storms, is rated at stop 8, “More auroras from less solar input?” (False) and in our check of the May 2024 aurora, “Needs a superflare at full field” (False).

Our checks on space weather: When Earth’s magnetic field falters, section 5 · The Sun has two worst cases · The Carrington storm left no detectable radiation mark in the ice · Superflares and the Sun’s extreme particle storms · A 7,300-year-old solar storm · Aurora met the equatorial anomaly in May 2024 · “A ‘small’ flare?”, stop 1 · “More auroras from less solar input?”, stop 8

8 · Reading a claim about the field

Eight questions to ask of the next one

Each comes with an example from our checks, quoted from the page.

  1. 01

    Which quantity: the dipole, the field at one place, or a pole?

    They move at very different speeds: from 1900 to 2025 the dipole’s pole moved about 250 km and the north dip pole about 2,500 km [7 Table 3]. On our review of October 20: “The rate part holds: IGRF-14 models the largest fall in field strength over North America, −128 nT a year, and our grid agrees. But that is 0.27% a year in a field of 43,000–55,000 nT, about twice the South Atlantic minimum”.

  2. 02

    Over what span, and at what resolution?

    On our check of a 2026 paper: “the paper’s rate is a straight line fitted through 1,500–2,000 years of a model that resolves centuries; today’s is a 185-year trend in direct measurements, and the two are not the same quantity”.

  3. 03

    Does the cited source say it?

    Our “Misread” verdicts are for real sources read wrongly. On October 22: “The paper’s record spans 262.5–258.75 million years ago; its authors tie the instability to the end-Guadalupian extinction, ‘though the causation remains unproven’”. Where the source doesn’t say it at all, the verdict is Unsupported: on our check of the excursions paper, of “80 to 90% as bad as a reversal”, “No such figure or comparison appears in the paper.”

  4. 04

    Measured, or forecast and extrapolated?

    On our review of the forecasts: “The 12.6 nT a year in the chart is IGRF-14’s forecast pace for 2025–30, not a measured one.” Of the video’s pole flip in the 2040s (stop 10): “The extrapolated line is the presenter’s own curve, not a scientific forecast.”

  5. 05

    Is the rate compared like with like?

    On our review of the record, beside the video’s “we go from 5% loss per century to 5% per decade”: “The dipole lost about half a percent per decade between 2000 and 2020”, slower than around 1980.

  6. 06

    Has the pattern been tested against chance?

    On our test of the 6,000-year cycle: “A real 6,000-year cycle counted from today would leave the six best-dated excursions about 470 years from its marks, allowing for their dating errors. They sit 1,640 years away on average, which no simulated cycle produced in 200,000 tries”. On our out-of-sample test of a Sun–earthquake link: 4 of 9 large earthquakes inside the windows, “Consistent with chance.”

  7. 07

    Peer-reviewed paper, preprint or conference abstract?

    On our check of a mortality paper: “This item isn’t one of them: it is a preprint on SSRN, not peer reviewed”. Abstracts and preprints can be right, but they haven’t been checked by other scientists yet.

  8. 08

    Who benefits?

    The episodes end with promotions, which our pages quote and don’t rate. On October 24: “A paid service that scores locations, ‘Havenbound — A Safety Report for Any Coordinate on Earth’, sold by Havenbound LLC, a Wyoming limited liability company; not a source, and not rated here”, and “neither the description nor the presenter says whether S0 is paid by, or has a stake in, Havenbound”.

9 · Limits

What this guide can and can’t settle

What to watch for

  1. 01

    The next reference models: IGRF-14 runs to 2030 [7 p. 17], and the World Magnetic Model “must be updated regularly, typically every five years” [30 p. 4]. Their new forecasts will show whether the pace stays near 5% a century.

  2. 02

    The anomaly’s second lobe south-west of Africa, which ESA says has weakened “even faster” since 2020 [34].

  3. 03

    Swarm itself: its satellites “are all healthy”, and ESA hopes to “extend that record beyond 2030” [34].

  4. 04

    NOAA’s monitors at L1: SOLAR-1 has been the primary one since June 10, 2026 [94]; DSCOVR was taken offline in July 2025, and we found no later notice of its return [95]. The last hour of warning before a storm depends on them [93].

  5. 05

    What the May 2024 measurements show about transformer heating when scaled up to March 1989 or to the TPL-007 benchmark, which NERC has asked industry and manufacturers to assess [80 p. 15].

Code and numbers

Run them

All four need only Python 3. Put them in one folder and run, for example, python3 field_decline.py. The first time, they download their inputs: the IGRF-14 coefficients and the WMM2025 package from NOAA [2][29]; the PADM2M and GGF100k files from the EarthRef archive [3][23]; the Oulu neutron monitor’s data [99]; and GFZ’s Kp file, 5.5 MB [60]. They check each file against the one we used, where it does not change daily, and read the archives’ data files only. Each prints its results and writes a JSON file; python3 claim6_scalings.py --grid adds the global grid of cutoffs, which takes a long time.

10 · The checks behind this guide

Every check this guide draws on

Each with its kind and what it found, in its own words.

The record

Today’s field

Timescales and the coming excursion

Life under a weak field

What physics rules out

Space weather

Sources

Bracketed numbers refer to this list. For the S0 News episodes and the video, times are positions in the video (hours:minutes:seconds, or minutes:seconds for the video). For papers and reports, p. and pp. give the printed page, § a section, Table a table, and Abstract, Teaser, Introduction, Results, Discussion, Plain Language Summary or Attachment the part we read; for the UNSCEAR annexes, ¶ gives a paragraph. Licences are given where we quote or adapt; data files and model files are charted as our own analysis, with credit. Links checked October 26, 2026.

  1. 1

    Finlay, C. C., Aubert, J., and Gillet, N. (2016). Gyre-driven decay of the Earth’s magnetic dipole. Nature Communications 7, 10422. CC BY 4.0.

  2. 2

    IAGA Working Group V-MOD (2024). International Geomagnetic Reference Field, 14th generation (IGRF-14): Gauss coefficients 1900–2025 and secular variation 2025–2030. Data file, NOAA National Centers for Environmental Information; retrieved October 26, 2026, the same file, by its SHA-256, as on our earlier pages.

  3. 3

    Ziegler, L. B., Constable, C. G., Johnson, C. L., and Tauxe, L. (2011). PADM2M: a penalized maximum likelihood model of the 0–2 Ma palaeomagnetic axial dipole moment. Geophysical Journal International 184(3), 1069–1089. Model file from the EarthRef Digital Archive, ERDA 1138; the paper’s publisher page showed a bot check, so we used the model file only.

  4. 4

    Singer, B. S., Jicha, B. R., Mochizuki, N., and Coe, R. S. (2019). Synchronizing volcanic, sedimentary, and ice core records of Earth’s last magnetic polarity reversal. Science Advances 5(8), eaaw4621. CC BY-NC 4.0; full text read at Europe PMC.

  5. 5

    SpaceWeatherNews (S0) (October 24, 2026). Earth Field Chaos, Solar Storm Deaths | S0 News Oct.24.2026. YouTube video. Quoted from the automatically generated transcript; times are positions in the video. Reviewed on our page for the episode.

  6. 6

    SpaceWeatherNews (S0) (2026). THE DISASTER CYCLE – The Evidence. YouTube video, quoted from the transcript behind our check of the video; times are positions in the video.

  7. 7

    Beggan, C. D., Kloss, C., Amblard, P., et al. (2026). International geomagnetic reference field: the fourteenth generation. Earth, Planets and Space 78, 127. CC BY 4.0.

  8. 8

    Channell, J. E. T. (2012). Geomagnetic excursions in the Brunhes and Matuyama Chrons: Do they come in bunches? Geophysical Research Abstracts 14, EGU2012-10478. Conference abstract, EGU General Assembly 2012.

  9. 9

    SpaceWeatherNews (S0) (October 22, 2026). Magnetic Pole Shift & The Largest Extinction Ever | S0 News Oct.22.2026. YouTube video. Quoted from the automatically generated transcript; times are positions in the video. Reviewed on our page for the episode.

  10. 10

    SpaceWeatherNews (S0) (October 23, 2026). Rapid Pole Shifts Confirmed, Micro-Disruptor | S0 News Oct.23.2026. YouTube video. Quoted from the automatically generated transcript; times are positions in the video. Reviewed on our page for the episode.

  11. 11

    Korte, M., Brown, M. C., Panovska, S., and Wardinski, I. (2019). Robust characteristics of the Laschamp and Mono Lake geomagnetic excursions: results from global field models. Frontiers in Earth Science 7, 86. CC BY 4.0.

  12. 12

    Glassmeier, K.-H., and Vogt, J. (2010). Magnetic polarity transitions and biospheric effects. Space Science Reviews 155(1–4), 387–410. Abstract only.

  13. 13

    British Geological Survey. Reversals: Magnetic Flip. Education page, read October 26, 2026.

  14. 14

    United Nations Scientific Committee on the Effects of Atomic Radiation (2000). Sources and Effects of Ionizing Radiation, Annex B: Exposures from natural radiation sources.

  15. 15

    United Nations Scientific Committee on the Effects of Atomic Radiation (2024). Annex B: Evaluation of public exposure to ionizing radiation.

  16. 16

    Davies, J. H., and Davies, D. R. (2010). Earth’s surface heat flux. Solid Earth 1(1), 5–24. CC BY 3.0. Abstract only.

  17. 17

    Fu, H., Zhang, S., Condon, D. J., and Xian, H. (2022). Secular change of true polar wander over the past billion years. Science Advances 8(41), eabo2753; full text read at Europe PMC.

  18. 18

    Kovaltsov, G. A., Mishev, A., and Usoskin, I. G. (2012). A new model of cosmogenic production of radiocarbon 14C in the atmosphere. Earth and Planetary Science Letters 337–338, 114–120. We read the arXiv version and re-ran its model.

  19. 19

    Glassmeier, K.-H., Vogt, J., Stadelmann, A., and Buchert, S. (2004). Concerning long-term geomagnetic variations and space climatology. Annales Geophysicae 22, 3669–3677. CC BY 3.0.

  20. 20

    NOAA Space Weather Prediction Center (2023). NOAA Space Weather Scales, with the table of the scales dated December 11, 2023. Read October 26, 2026.

  21. 21

    Maffei, S., Livermore, P. W., and Mound, J. E. (2025). Sensitivity of the auroral zones to temporal changes in Earth’s internal magnetic field. Journal of Space Weather and Space Climate 15, 29. CC BY 4.0; excerpts read at the publisher’s site.

  22. 22

    Olson, P., and Amit, H. (2015). Mantle superplumes induce geomagnetic superchrons. Frontiers in Earth Science 3, 38. CC BY 4.0.

  23. 23

    Panovska, S., Constable, C. G., and Korte, M. (2018). GGF100k model files, with the dipole and axial dipole moments every 50 years. EarthRef Digital Archive, ERDA 2382.

  24. 24

    Panovska, S., Constable, C. G., and Korte, M. (2018). Extending global continuous geomagnetic field reconstructions on timescales beyond human civilization. Geochemistry, Geophysics, Geosystems 19(12), 4757–4772. We read the copy in GFZpublic.

  25. 25

    Shinu, S., Mahgoub, A. N., and Korte, M. (2026). Rates of geomagnetic field change during excursions and reversals. Journal of Geophysical Research: Solid Earth 131(10), e2025JB033422. CC BY 4.0. As read for our check of the paper.

  26. 26

    Laj, C., and Kissel, C. (2015). An impending geomagnetic transition? Hints from the past. Frontiers in Earth Science 3, 61. CC BY 4.0.

  27. 27

    Nilsson, A., Suttie, N., Stoner, J. S., and Muscheler, R. (2022). Recurrent ancient geomagnetic field anomalies shed light on future evolution of the South Atlantic Anomaly. PNAS 119(24), e2200749119. CC BY-NC-ND 4.0, quoted only; full text read at Europe PMC.

  28. 28

    Chulliat, A., Brown, W., Nair, M., et al. (2025). The US/UK World Magnetic Model for 2025–2030: Technical Report. NOAA National Centers for Environmental Information. CC0; we read the copy in NOAA’s library.

  29. 29

    NOAA National Centers for Environmental Information and British Geological Survey (2024). World Magnetic Model 2025: coefficients and test values (WMM2025COF.zip). Data file.

  30. 30

    NOAA National Centers for Environmental Information and British Geological Survey (2025). State of the Geomagnetic Field, December 2025. CC0.

  31. 31

    NOAA National Centers for Environmental Information (2026, January 9). 2025 WMM Annual Report is Released. News item.

  32. 32

    European Space Agency (2020, May 20). Swarm probes weakening of Earth’s magnetic field. News release.

  33. 33

    Finlay, C. C., Kloss, C., and Gillet, N. (2025). Core field changes from eleven years of Swarm satellite observations. Physics of the Earth and Planetary Interiors 368, 107447. CC BY 4.0. Abstract, as in the record at DTU Orbit.

  34. 34

    European Space Agency (2025, October 13). Swarm reveals growing weak spot in Earth’s magnetic field. News release.

  35. 35

    Finlay, C. C., Kloss, C., Olsen, N., Hammer, M. D., Tøffner-Clausen, L., Grayver, A., and Kuvshinov, A. (2020). The CHAOS-7 geomagnetic field model and observed changes in the South Atlantic Anomaly. Earth, Planets and Space 72, 156. CC BY 4.0.

  36. 36

    Space Telescope Science Institute. HST Cycle 26 Primer: Orbital Constraints, section “South Atlantic Anomaly (SAA)”. Read October 26, 2026.

  37. 37

    NASA (2020, August 17). NASA researchers track slowly splitting ‘dent’ in Earth’s magnetic field.

  38. 38

    Tarduno, J. A., Watkeys, M. K., Huffman, T. N., et al. (2015). Antiquity of the South Atlantic Anomaly and evidence for top-down control on the geodynamo. Nature Communications 6, 7865. CC BY-NC-SA 4.0, quoted only.

  39. 39

    Engbers, Y. A., Grappone, J. M., Mark, D. F., and Biggin, A. J. (2022). Low paleointensities and Ar/Ar ages from Saint Helena provide evidence for recurring magnetic field weaknesses in the South Atlantic. Journal of Geophysical Research: Solid Earth 127(3), e2021JB023358. CC BY 4.0. Abstract only.

  40. 40

    Engbers, Y. A., Biggin, A. J., and Bono, R. K. (2020). Elevated paleomagnetic dispersion at Saint Helena suggests long-lived anomalous behavior in the South Atlantic. PNAS 117(31), 18258–18263. Its significance statement, as deposited with Crossref.

  41. 41

    Pavón-Carrasco, F. J., and De Santis, A. (2016). The South Atlantic Anomaly: The key for a possible geomagnetic reversal. Frontiers in Earth Science 4, 40. CC BY 4.0.

  42. 42

    Clement, B. M. (2004). Dependence of the duration of geomagnetic polarity reversals on site latitude. Nature 428, 637–640. Abstract only.

  43. 43

    Camps, P., Coe, R. S., and Prévot, M. (1999). Transitional geomagnetic impulse hypothesis: geomagnetic fact or rock-magnetic artifact? Journal of Geophysical Research: Solid Earth 104(B8), 17747–17758. Abstract only.

  44. 44

    Bogue, S. W., and Glen, J. M. G. (2010). Very rapid geomagnetic field change recorded by the partial remagnetization of a lava flow. Geophysical Research Letters 37(21). Abstract only.

  45. 45

    Sagnotti, L., Giaccio, B., Liddicoat, J., Nomade, S., Renne, P., Scardia, G., and Sprain, C. (2016). How fast was the last geomagnetic reversal? Experimental constraints from the Sulmona lacustrine sequence (central Apennines, Italy). EGU General Assembly 2016, EGU2016-3851. Conference abstract, CC BY 3.0, summarizing the authors’ papers in Geophysical Journal International 199, 1110–1124 (2014) and 204, 798–812 (2016).

  46. 46

    Evans, M. E., and Muxworthy, A. R. (2018). A re-appraisal of the proposed rapid Matuyama–Brunhes geomagnetic reversal in the Sulmona Basin, Italy. Geophysical Journal International 213(3), 1744–1750. The authors’ abstract, in Imperial College’s repository.

  47. 47

    Sagnotti, L., Giaccio, B., Liddicoat, J., Caricchi, C., Nomade, S., and Renne, P. (2019). On the reliability of the Matuyama–Brunhes record in the Sulmona Basin—Comment to ‘A reappraisal of the proposed rapid Matuyama–Brunhes geomagnetic reversal in the Sulmona Basin, Italy’ by Evans and Muxworthy (2018). Geophysical Journal International 216(1), 296–301. Title and record only.

  48. 48

    Davies, C. J., and Constable, C. G. (2020). Rapid geomagnetic changes inferred from Earth observations and numerical simulations. Nature Communications 11, 3371. CC BY 4.0. Abstract only.

  49. 49

    Wang, R. J., Al-Saffar, S. I., Rogers, J., and Hahn, M. W. (2023). Human generation times across the past 250,000 years. Science Advances 9(1), eabm7047. Abstract only.

  50. 50

    Sato, T. (2016). Evaluation of world population-weighted effective dose due to cosmic ray exposure. Scientific Reports 6, 33932. CC BY 4.0.

  51. 51

    Larsen, N., Usoskin, I., Mishev, A., Koldobskiy, S., and Väisänen, P. (2026). Reduced geomagnetic shielding during the Laschamps excursion and its impact on cosmic-ray-induced atmospheric radiation. Journal of Geophysical Research: Space Physics 131(2), e2025JA034820. CC BY 4.0; we read the copy in the University of Oulu’s repository.

  52. 52

    University of Oulu (2026, March 25). If the Laschamps geomagnetic excursion happened today, aviation radiation exposure would be radically altered – with “shielded pockets” in the north. News release.

  53. 53

    Simon, Q., Thouveny, N., Bourlès, D. L., et al. (2016). Authigenic 10Be/9Be ratio signatures of the cosmogenic nuclide production linked to geomagnetic dipole moment variation since the Brunhes/Matuyama boundary. Journal of Geophysical Research: Solid Earth 121(11), 7716–7741. CC BY-NC-ND 4.0. Abstract only.

  54. 54

    Cauquoin, A., Raisbeck, G. M., Jouzel, J., and Paillard, D. (2014). Use of 10Be to predict atmospheric 14C variations during the Laschamp excursion: high sensitivity to cosmogenic isotope production calculations. Radiocarbon 56(1), 67–82. Abstract only.

  55. 55

    Cooper, A., Turney, C. S. M., Palmer, J., et al. (2021). A global environmental crisis 42,000 years ago. Science 371(6531), 811–818. As read for our review of the October 22 episode.

  56. 56

    Picin, A., Benazzi, S., Blasco, R., et al. (2021). Comment on “A global environmental crisis 42,000 years ago”. Science 374(6570), eabi8330. Abstract only, as deposited with Crossref.

  57. 57

    European Southern Observatory (2022, April 20). Astronomers discover micronovae, a new kind of stellar explosion. Press release eso2207.

  58. 58

    Scaringi, S., Groot, P. J., Knigge, C., et al. (2022). Localized thermonuclear bursts from accreting magnetic white dwarfs. Nature 604, 447–450. Abstract only.

  59. 59

    Lloyd’s and Atmospheric and Environmental Research (2013). Solar storm risk to the North American electric grid. Lloyd’s, London.

  60. 60

    Matzka, J., Stolle, C., Yamazaki, Y., Bronkalla, O., and Morschhauser, A. (2021). The geomagnetic Kp index and derived indices of geomagnetic activity. Space Weather 19, e2020SW002641. Data: Matzka, J., Bronkalla, O., Tornow, K., Elger, K., and Stolle, C. (2021). Geomagnetic Kp index, V. 1.0, GFZ Data Services, as of October 25, 2026. Kp from Geomagnetic Observatory Niemegk, GFZ Helmholtz Centre for Geosciences, CC BY 4.0; counted and ranked by us.

  61. 61

    Cliver, E. W., and Dietrich, W. F. (2013). The 1859 space weather event revisited: limits of extreme activity. Journal of Space Weather and Space Climate 3, A31.

  62. 62

    Green, J. L., and Boardsen, S. (2006). Duration and extent of the great auroral storm of 1859. Advances in Space Research 38(2), 130–135.

  63. 63

    Hayakawa, H., Ebihara, Y., Hand, D. P., et al. (2018). Low-latitude aurorae during the extreme space weather events in 1859. The Astrophysical Journal 869(1), 57.

  64. 64

    Hayakawa, H., Iwahashi, K., Tamazawa, H., et al. (2016). East Asian observations of low-latitude aurora during the Carrington magnetic storm. Publications of the Astronomical Society of Japan 68(6), 99.

  65. 65

    Love, J. J., Hayakawa, H., and Cliver, E. W. (2019). Intensity and impact of the New York Railroad superstorm of May 1921. Space Weather 17(8), 1281–1292. CC BY 4.0.

  66. 66

    Silverman, S. M., and Cliver, E. W. (2001). Low-latitude auroras: the magnetic storm of 14–15 May 1921. Journal of Atmospheric and Solar-Terrestrial Physics 63(5), 523–535.

  67. 67

    Love, J. J., Lucas, G. M., Kelbert, A., et al. (2025). The impact of the May 1921 superstorm on American telecommunication systems. Space Weather 23(7), e2025SW004563. CC BY 4.0.

  68. 68

    Koči, E., and Valach, F. (2023). The record of the magnetic storm on 15 May 1921 in Stará Ďala (present-day Hurbanovo) and its compliance with the global picture of this extreme event. Annales Geophysicae 41(2), 355–368. CC BY 4.0.

  69. 69

    NOAA National Environmental Satellite, Data, and Information Service (2015, October 19). When solar storms attack: Space weather and our infrastructure.

  70. 70

    Hydro-Québec. Understanding electricity: March 1989 blackout. Read October 26, 2026.

  71. 71

    North American Electric Reliability Corporation (2012, February). 2012 Special Reliability Assessment Interim Report: Effects of Geomagnetic Disturbances on the Bulk Power System.

  72. 72

    Wik, M., Pirjola, R., Lundstedt, H., et al. (2009). Space weather events in July 1982 and October 2003 and the effects of geomagnetically induced currents on Swedish technical systems. Annales Geophysicae 27(4), 1775–1787. CC BY 3.0.

  73. 73

    Pulkkinen, A., Lindahl, S., Viljanen, A., and Pirjola, R. (2005). Geomagnetic storm of 29–31 October 2003: Geomagnetically induced currents and their relation to problems in the Swedish high-voltage power transmission system. Space Weather 3(8), S08C03.

  74. 74

    Gaunt, C. T. (2014). Reducing uncertainty – responses for electricity utilities to severe solar storms. Journal of Space Weather and Space Climate 4, A01.

  75. 75

    Gaunt, C. T., and Coetzee, G. (2007). Transformer failures in regions incorrectly considered to have low GIC-risk. 2007 IEEE Lausanne Power Tech, 807–812.

  76. 76

    NOAA National Weather Service (2004). Service Assessment: Intense Space Weather Storms October 19 – November 07, 2003. US Department of Commerce.

  77. 77

    Baker, D. N., Li, X., Pulkkinen, A., et al. (2013). A major solar eruptive event in July 2012: Defining extreme space weather scenarios. Space Weather 11, 585–591. Abstract only.

  78. 78

    SpaceX (2022). Updates: “Geomagnetic storm and recently deployed Starlink satellites” (February 8) and “SpaceX’s approach to space sustainability and safety” (February 22). Read October 26, 2026.

  79. 79

    Fang, T.-W., Kubaryk, A., Goldstein, D., et al. (2022). Space weather environment during the SpaceX Starlink satellite loss in February 2022. Space Weather 20, e2022SW003193. CC BY 4.0. Abstract only.

  80. 80

    North American Electric Reliability Corporation (2026, January). May 2024 GMD Event Review. NERC staff report.

  81. 81

    World Data Center for Geomagnetism, Kyoto, Nose, M., Iyemori, T., Sugiura, M., and Kamei, T. (2015). Geomagnetic Dst index. Final values for March 1989 and October 2003 (the latter as on our review of the record); provisional values for February 2022 and May 2024. Read October 26, 2026.

  82. 82

    Riley, P. (2012). On the probability of occurrence of extreme space weather events. Space Weather 10, S02012. Abstract only.

  83. 83

    Love, J. J. (2012). Credible occurrence probabilities for extreme geophysical events: Earthquakes, volcanic eruptions, magnetic storms. Geophysical Research Letters 39, L10301. Abstract only; the chance per decade and its range are from the author’s slides presenting the paper, hosted by the National Academies.

  84. 84

    Kataoka, R. (2013). Probability of occurrence of extreme magnetic storms. Space Weather 11, 214–218. Abstract only.

  85. 85

    Love, J. J., Rigler, E. J., Pulkkinen, A., and Riley, P. (2015). On the lognormality of historical magnetic storm intensity statistics: Implications for extreme-event probabilities. Geophysical Research Letters 42, 6544–6553. Abstract only.

  86. 86

    Riley, P., and Love, J. J. (2017). Extreme geomagnetic storms: Probabilistic forecasts and their uncertainties. Space Weather 15, 53–64. Abstract only.

  87. 87

    Moriña, D., Serra, I., Puig, P., and Corral, Á. (2019). Probability estimation of a Carrington-like geomagnetic storm. Scientific Reports 9, 2393. CC BY 4.0.

  88. 88

    Chapman, S. C., Horne, R. B., and Watkins, N. W. (2020). Using the aa index over the last 14 solar cycles to characterize extreme geomagnetic activity. Geophysical Research Letters 47, e2019GL086524. CC BY 4.0; we read the copy in NORA.

  89. 89

    North American Electric Reliability Corporation (2016, May 12). Benchmark Geomagnetic Disturbance Event Description. White paper.

  90. 90

    HM Government (2025). National Risk Register, 2025 edition. Crown copyright.

  91. 91

    Eastwood, J. P., Biffis, E., Hapgood, M. A., et al. (2017). The economic impact of space weather: Where do we stand? Risk Analysis 37, 206–218. Abstract only.

  92. 92

    National Research Council (2008). Severe Space Weather Events: Understanding Societal and Economic Impacts: A Workshop Report. The National Academies Press. Pages read as images at the publisher’s site.

  93. 93

    NOAA Space Weather Prediction Center. Coronal Mass Ejections. Read October 26, 2026.

  94. 94

    NOAA Office of Satellite and Product Operations. SOLAR-1 (SWFO-L1) status: launched September 24, 2025; “Operational Date: 6/10/2026”; “Primary”. Read October 26, 2026.

  95. 95

    NOAA Office of Satellite and Product Operations (2025). DSCOVR notices of July 15 and July 16, 2025: after “a software bus anomaly”, DSCOVR is “offline until further notice”.

  96. 96

    North American Electric Reliability Corporation. EOP-010-1: Geomagnetic Disturbance Operations. Reliability standard.

  97. 97

    North American Electric Reliability Corporation. TPL-007-4: Transmission System Planned Performance for Geomagnetic Disturbance Events. Reliability standard.

  98. 98

    Herbst, K., Kopp, A., and Heber, B. (2013). Influence of the terrestrial magnetic field geometry on the cutoff rigidity of cosmic ray particles. Annales Geophysicae 31, 1637–1643. CC BY 3.0.

  99. 99

    Oulu Cosmic Ray Station, Sodankylä Geophysical Observatory, University of Oulu. Oulu neutron monitor: corrected count rates, yearly 1965–2025, monthly 1965–2026 and hourly for October–November 2003 and May 2024, from the station’s online query, October 26, 2026; analysed by us. We acknowledge the station and the Sodankylä Geophysical Observatory, as the station asks.

  100. 100

    Bruno, A., Martucci, M., Cafagna, F. S., et al. (2021). Solar-cycle variations of South-Atlantic Anomaly proton intensities measured with the PAMELA mission. arXiv:2108.06407, the version we read; its p. 7 gives the solar modulation at the Tsumeb neutron monitor, Namibia, from the monitor’s count rates for 2006–2014.

  101. 101

    Gvozdevsky, B., Dorman, L., Belov, A., Gushchina, R., Eroshenko, E., Preobrazhensky, M., and Yanke, V. (2017). The secular variations of cosmic ray cutoff rigidities, caused by century variations in geomagnetic field, and cosmic ray variations. 35th International Cosmic Ray Conference, PoS(ICRC2017)067. CC BY-NC-ND 4.0.

Method

How this was checked

On October 26, 2026, we checked every statement, number and quotation in this guide against its source, 101 sources in all: the automatically generated transcripts of three S0 News episodes and the transcript behind our check of the video; the IGRF-14 coefficients and paper, the World Magnetic Model’s coefficients, report and first-year review, and ESA’s and NASA’s pages on Swarm and the anomaly; the PADM2M and GGF100k model files and their papers; NOAA’s scales, GFZ’s Kp record, Kyoto’s Dst index and the Oulu neutron monitor’s data; the reports of NOAA, NERC, the National Research Council, Lloyd’s and the UK government on storms and their impacts; and papers and reviews on the anomaly’s history, the timescales of excursions and reversals, cosmic rays and doses, extinctions, polar wander, micronovae and the odds of a great storm, read in full where open and in abstract where not, as the source list says. DOIs were resolved through Crossref. The guide draws on 33 of the site’s own pages and 14 items of its video check, and reuses their verdicts and wording. A fresh reviewer then checked the built page against the sources, re-ran our scripts on the same inputs and resolved every DOI and link; the review found that our re-run of the radiocarbon model had the wrong geomagnetic cutoff function, which put claim 6’s production change at about 5% instead of about 4%, so we corrected it and re-ran everything, and it led us to take the video’s line out of claim 6, show all four neutron-monitor stations, date the anomaly’s second low by its depth, add the primary sources of the storm table and correct several citations and wordings. The verdicts did not change.

The verdicts and the items tagged “Our analysis” are our own assessment, built on the cited sources. Four of the six verdicts repeat the site’s existing checks: no 6,000-year cycle; no excursion under way; no extinction record; and no 90° turn of the crust. The first, that the field is weakening, states what the site’s pages have always said. The sixth, that a weaker field leaves Earth more exposed to solar storms, is new here: right in direction, overstated in size.

We computed, from the public files: the dipole moment, its tilt and pole at every IGRF-14 epoch from 1900 to 2025 and the 2030 forecast, its pace over every five-year span and longer span, and whether it is speeding up; the South Atlantic Anomaly’s minimum, its drift, its area below several levels and how deep its second low sits below the saddle joining it to the main anomaly; today’s dipole against the averages of PADM2M and GGF100k; and an upper bound for the extra cosmic-ray dose at the ground if the field vanished. For claim 6 we computed the geomagnetic latitude and cosmic-ray cutoff of 35 places in 1900 and 2025, the shift of the auroral oval and the magnetopause under published scalings, and global radiocarbon production in a re-run of Kovaltsov and colleagues’ model, which reproduces the paper’s own checks; from GFZ’s Kp record since 1932, how often each storm level is reached; and, from the Oulu neutron monitor, the swings over each solar cycle and in two storms. The storm odds marked as our conversion turn a published rate or return period into a chance per decade. Our field code reproduces NOAA’s official test values to 0.05 nT, the IGRF-14 paper’s 2025 minimum to the nanotesla and the Swarm team’s 0.9% growth since 2014. Some sources could not be read in full: the PADM2M paper’s publisher showed a bot check, so we used its model file, and several papers on storm odds and impacts were read only in their abstracts.

This page was researched and written with Claude, an AI assistant made by Anthropic, and then re-checked against its sources in a separate verification pass.

The verdicts on this page

False
The evidence contradicts the claim.
Overstated
Has some basis but claims much more than the evidence supports.
True
The evidence supports the claim.

All ten verdicts, and how we check

Corrections

Corrections and changes

If we get something wrong, we fix it and record the change below with the date. If a correction changes a verdict, the log says what the old verdict was.

Change log

  1. Published on factchecked.science after a full check against the sources. The guide brings together the site’s checks on Earth’s magnetic field, its excursions, the South Atlantic Anomaly and space weather, and adds our own runs of the IGRF-14, WMM2025, PADM2M and GGF100k model files, counts of storm days in GFZ’s Kp record, the scalings behind claim 6 and a reader’s checklist. After a fresh review the same day, we corrected the radiocarbon model behind claim 6, so production now rises about 4% since 1900, not 5%; took the video’s line out of claim 6; showed all four neutron-monitor stations in its chart; dated the anomaly’s second low by its depth; added the primary sources of the storm table; and fixed several citations and wordings.

Suggest a correction

Spotted an error, or know a better source? Tell us here. Submissions aren’t published.