South Atlantic Anomaly
Swarm satellites have observed a weak-field region expanding from 2014 to 2025. This is a recent observational context, not evidence that a reversal is imminent.
ESA Swarm analysis ↗GEOMAGNETISM / 01
The poles in motion.
Explore modeled magnetic-pole movement from the Holocene to today. Choose a source and epoch, then zoom and pan the map.
436
NOAA YEARS
MODELED
MODELED
NOAA modeled dip-pole positions. Select a year to see the two poles.
Explore the last reversal ↓Loading data…
01 / MAP
Global trajectories
● MODELED POSITIONS
EQUIRECTANGULAR PROJECTION
NORTH POLE
SOUTH POLE
FULL TRAJECTORY
GEOMAGNETIC POLE
SURVEYED POLE
——Computed in your browser from the model coefficients on a 1° grid at sea level.
GEOMAGIA50 records identified in MagIC with explicit calendar ages; BP/ka ages are excluded. This subset does not represent the entire database.
Blue circles: result counts per 5° × 5° cell, within ±250 years of the selected epoch using central ages. Age uncertainties are not propagated. Counts are not density per unit area and do not enter pole or field calculations.
GEOMAGIA50 · Brown et al. (2015) ↗
MagIC · CC BY 4.0 ↗
Source and scope ↗
Direct determinations of the dip poles by ground parties, aircraft and one ship, 1831–2007. Each is a mean or quiet-level position: the dip pole wanders tens of kilometres around it every day.
Hollow diamonds mark the surveys; the distance compares each survey with the modeled position of the nearest model year. Surveys are independent of the models and do not enter their calculation.
Newitt et al. (2009) ↗
Mandea & Dormy (2003) ↗
Barton (2002) ↗
Ross (1834) ↗
03 / NOAA · 1590–2025
Travel through time
SELECTED YEAR2025
04 / CHANGE
Latitude through the centuries
Angular distance from the equator, in degrees. Both series use absolute latitude for comparison.
North South
05 / DRIFT RATE
How fast do the poles move?
Great-circle distance since the previous model epoch divided by elapsed years. CALS10k.2 values average 50 years (40 for the last interval).
North South
06 / GLOBAL DIPOLE
Field strength and pole drift
Dipole moment from independent field-model coefficients. NOAA pole positions and field intensity are separate model products; the GUFM1/IGRF boundary is marked at 1900, and the shaded GUFM1 years before 1840 rely on an extrapolated g₁⁰.
SELECTED MOMENT—× 10²² A·m²—
07 / PARALLEL RECORDS
Independent records beside the model
These observations come from other instruments and archives. They share a time axis with the selected series, but none of them is used to calculate the poles or the field.
Data unavailable.
Compass readings in London
Declination calculated at London from GUFM1 and IGRF-14, with three early measurements. Gellibrand's comparison of these readings in 1634 revealed that declination changes with time.
Early readings are rounded and come from different London sites; they are not corrected to the model location. Gunter’s 1622 marker is ≈6° E, rounded from his Limehouse table.
Gunter (1673), p. 279 ↗Sunspots and the Maunder Minimum
Annual mean sunspot number, SILSO version 2.0, from 1749. Solar activity drives the external field and aurorae; it does not move the core field's dip poles.
Length of day and geomagnetic jerks
Annual mean excess length of day from IERS, in milliseconds. Decadal changes are largely attributed to exchanges of angular momentum between core and mantle; jerks are abrupt changes in the field's secular variation.
Great aurorae on the map
Select a storm to show it on the map. The dashed line is the authors' equatorward auroral boundary, drawn here with the dipole of the same year; the band shows the usual auroral zone near 65–70° geomagnetic latitude.
Radiocarbon in the atmosphere
IntCal20 Δ¹⁴C, in ‰. A weaker dipole lets more cosmic rays produce ¹⁴C, but solar activity and the carbon cycle also shape this curve. Marked years are extreme solar particle events found in tree rings.
08 / DEEP TIME · GGFMB
The last magnetic reversal
A separate reconstruction of the global magnetic field, 900,000–700,000 years ago. Explore the signed axial dipole as the field changed polarity.
MODEL AGE780 ka BP
POLARITY—
AXIAL DIPOLE g₁⁰—µT
DIPOLE MOMENT—× 10²² A·m²
09 / FIELD STRENGTH
Dipole moment through the reversal
The same GGFMB epochs, in 10²² A·m². A weak field accompanies the polarity transition.
The shaded 799–770 ka interval is the authors' approximate reversal window. GGFMB is a model from sediment records; during transition the field is complex, so a single north/south pole path would be misleading.
Independent evidence: beryllium-10
Marine sediment records show increased cosmogenic ¹⁰Be production around the Matuyama–Brunhes transition as the dipole weakened. Their age models and sampling differ from GGFMB, so no point-by-point correlation is implied here.
Read the sediment study ↗10 / POLARITY TIMESCALE
Reversals over 3.6 million years
Normal (light) and reversed (dark) polarity from the GTS2012 timescale, with two dated excursions. The bracket marks the 900–700 ka GGFMB window above.
Data unavailable.
11 / OBSERVATIONS
Independent clues
Read these comparisons carefully
The curves use different models, timescales and resolutions. A pole's drift rate, a regional weak-field patch and a global polarity reversal describe different aspects of the magnetic field.
SCIENTIFIC NOTE
What does this map show?
Magnetic dip poles are where Earth's magnetic field is vertical. NOAA supplies annual modeled positions for 1590–2025; CALS10k.2 is a separate Holocene reconstruction from 8000 BCE.
For CALS10k.2, cubic splines define spherical-harmonic coefficients through time. We solve for zero north and east field components at sea level. Lines join model epochs, not annual observations.
Predictive modeled magnetic dip poles, separate from the historical series.
WMM2025 ↗ North SouthBEYOND 1590
A deeper magnetic past
CALS10k.2 is available in the separate historical view. Other models reach farther back, but their pole paths have not been computed here.
CALS10k.2~10,000 years · GFZ model and coefficients
ArchKalmag14k~14,000 years · global model, 50-year steps
SHA.DIF.14k~14,000 years · modeled north dip-pole path
GGF100k~100,000 years · long-term field reconstruction
CALS10k.2 and NOAA are distinct models and give different positions in 1590. Their paths are separate; uncertainty grows farther into the past.
ABOUT · CREDITS
An independent atlas of published models
This atlas is an independent educational visualization. It redraws published magnetic-field models and surveys with their own numbers, in your browser, and does not produce new scientific results. Every value shown is a model result or a published measurement, with its source linked where it appears.
The atlas itself makes no network request at runtime once its files are loaded, and share links carry only the selected view. Audience measurement (Google Analytics) runs only if you accept it; the choice is stored on this device.
- NOAA NCEI — modeled dip poles 1590–2025, IGRF-14 and WMM2025 test values · U.S. Government work, public domain · ncei.noaa.gov ↗
- WMM2025 — predictive model 2025–2030, NOAA NCEI and British Geological Survey · ncei.noaa.gov ↗
- CALS10k.2 — Holocene field model, Constable, Korte & Panovska (2016), EarthRef ERDA 2207 · CC BY 4.0 · earthref.org ↗
- GGFMB — Matuyama–Brunhes reversal model, Mahgoub et al. (2023), EarthRef ERDA 2548 · CC BY 4.0 · earthref.org ↗
- GUFM1 — historical field model 1590–1990, Jackson, Jonkers & Walker (2000) · doi:10.1098/rsta.2000.0569 ↗
- GEOMAGIA50 / MagIC — archeomagnetic and volcanic database records, Brown et al. (2015); EarthRef CC BY 4.0 · doi:10.1186/s40623-015-0232-0 ↗
- IERS EOP 20 C04 · SILSO sunspots (via NOAA SWPC) · IntCal20 — independent context series, each cited in its section
- NASA Blue Marble — U.S. Government work, public domain · Natural Earth / world-atlas — public domain (data) and ISC license (topology)
- Noto — Google, SIL Open Font License 1.1
Report an error or a missing source ↗Source code and data scripts ↗