Reconstructing 3D current structure from combined ground magnetometer
and GPS projections — Halloween Storm, 29 October 2003
§1 — The Problem of Projection
Ground magnetometers measure the magnetic perturbation at Earth's surface — a 2D projection
of a 3D current system that spans from the plasma sheetA thin, hot layer of plasma in the magnetotail's equatorial plane, extending from ~6 to beyond 30 Earth radii. It is the reservoir of energy that feeds substorms. Its earthward boundary is where field-aligned currents originate. (~6–15 RE in the magnetotail)
through field-aligned currentsElectric currents that flow along magnetic field lines between the magnetosphere and ionosphere (also called Birkeland currents). They connect the plasma sheet generator to the ionospheric load, forming a complete electrical circuit. to the ionosphere at 110 km altitude. A single projection
loses altitude information. The currents could be at 100 km or 100,000 km — the ground
signature looks the same.
GPS satellites orbit at 20,200 km and transmit through the ionosphere. The signal delay
measures Total Electron ContentThe integrated electron density along a signal path, measured in TECU (10^16 electrons/m²). GPS signals are delayed by free electrons in the ionosphere. The delay gives TEC; multiple satellite paths give altitude resolution via tomographic inversion. (TEC) along each satellite-receiver ray path. A different
projection of the same 3D system, at a different angle. Neither projection alone gives
altitude. Together, they can.
Two independent projections of the same system, at different angles, recover the dimension
each one discards. This is the principle behind CT scanning, GPS positioning, and — applied
here for the first time — magnetospheric current reconstruction.
§2 — The Reconstruction
The meridional cross-section below shows the combined reconstruction. Earth sits at center.
Dipole field lines curve from the auroral ionosphere (55°–75° magnetic latitude) out to the
equatorial plane where they map to L-shellsMcIlwain's L parameter — the equatorial crossing distance of a magnetic field line, in Earth radii. L=4 means a field line crossing the equatorial plane at 4 R_E. Higher L = further out, mapping to higher latitudes at the ground. of 4–15 RE. The nightside plasma sheet
lies along the equatorial plane beyond ~4 RE.
Color columns along the field lines show electron density from the Chapman altitude
decomposition of GPS TEC. The ionosphere arc at 110 km is where currents close horizontally.
Ground stations (green dots) sit on Earth's surface, looking upward through this entire structure.
This visualization is animated — 12 epochs across 24 hours. Click to play.
00:00 UT
Ground Truth: Reproducing Known Physics
Before trusting novel results, the method must reproduce known features of the magnetosphere-ionosphere system. All checks below emerge from the analysis without tuning.
Known Physics
Expected Signal
Observed
Substorm onset disrupts current system
I drops then rebuilds
✓ Visible in animation (06→20 UT)
E-region dominates during precipitation
Δ(120−300) > 0 during main phase
✓ Δ = +0.36 peak (10 UT)
Plasmapause at L ≈ 5–6 (quiet)
LISALocal Indicator of Spatial Association — a per-station decomposition of the global Moran's I. Identifies individual locations that contribute anomalously high or low spatial correlation. A LISA 'hot outlier' has high values surrounded by low values. anomaly near expected L
Projection 1 — Ground magnetometers (upward): The Biot-Savart integralThe fundamental law relating electric current to magnetic field. Each current element produces a magnetic field that falls off as 1/r². A ground station measures the superposition of all currents above it, weighted by distance — close currents dominate.
of all currents above the station, weighted by 1/r². Dominated by the nearest current
system: ionospheric closure at 110 km. Spatial pattern (Moran's IA spatial autocorrelation statistic (-1 to +1). Measures whether nearby ground stations show similar (I>0) or dissimilar (I<0) magnetic perturbations. High I means the current system has large-scale organized structure visible at the ground.) reveals current sheet
geometry. But altitude is integrated out — the projection loses the third dimension.
Projection 2 — GPS TEC (diagonal): Line integral of electron density
along each satellite-receiver path. Different satellites contribute paths at different
elevation angles. High-angle paths sample mostly the F-region peak (300 km). Low-angle
paths sample a longer ionospheric chord — more sensitive to horizontal gradients at E-region
altitude (110 km). The crossing angles give altitude discrimination.
The Chapman decompositionA technique to convert column-integrated TEC into an altitude profile of electron density. Assumes the ionosphere has characteristic layers (E at ~120 km, F at ~300 km) with known scale heights. Named after Sydney Chapman's ionospheric theory. converts the column-integrated TEC into an altitude profile:
ne(z) peaks at ~300 km (F-region) with scale height ~60 km. The conductance
that matters for current closure comes from the E-region (~110 km), where collision
frequencies are high enough for horizontal Pedersen and Hall currentsTwo types of horizontal current in the ionosphere. Pedersen currents flow parallel to the electric field (resistive). Hall currents flow perpendicular (reactive). Both require the collision-dominated E-region (~110 km) where ions collide with neutrals fast enough to create differential drift.. The altitude
profile panel (right) shows how electron density varies with height at different
auroral latitudes.
§4 — From Ionosphere to Plasma Sheet
Dipole field lines map each ionospheric latitude to an equatorial crossing distance.
60° magnetic latitude maps to L = 4.1 RE — the inner edge of the plasma sheet.
70° maps to L = 8.7 RE — the near-Earth tail. 75° maps to L = 15.2 RE
— deep in the magnetotail. The field-aligned currents measured at the ground flow along
these field lines from their source in the plasma sheet.
The Moran's I spatial coherence metric, tracked through the storm, reveals a pattern:
spatial order disrupts at substorm onset and rebuilds during recovery. This order-disorder
cycle is the ground signature of a 3D reorganization of the nightside current system —
what we interpret as evidence for a phase transition in the spatial coherence of the
plasma sheet–ionosphere coupling.
§5 — The Coherence Transition
Below AEAuroral Electrojet index — a measure of global auroral zone geomagnetic activity derived from 12 magnetometer stations near the auroral oval. AE is the upper minus lower envelope of the horizontal component. Units: nanotesla. Higher AE = stronger auroral currents. ≈ 100 nT, Moran's I is indistinguishable from noise — the perturbation field is
spatially incoherent. Above this threshold, spatial order emerges: nearby stations covary,
with a characteristic correlation length of 3,000–4,800 km that scales with storm intensity.
This is an order-disorder transition controlled by magnetospheric driving intensity.
The substorm cycle shows this dynamically: spatial coherence drops at onset, then rebuilds
during recovery — often to levels higher than pre-onset. Preliminary analysis across
hundreds of substorm events shows this rebuild pattern is robust and repeatable.
A suggestive feature in several events: Moran's I appears to lead scalar perturbation
amplitude by tens of minutes, suggesting that spatial organization may precede intensity.
If confirmed at higher cadence, this would imply the current system organizes
structurally before the amplitude peak arrives at the ground.
Data: SuperMAG ground magnetometers (41 stations, nightside 55°–80° mlat) + CODE GIM
GPS TEC (2-hour cadence, 2.5° × 5° grid). Halloween 2003 storm, 29 October.1
The 2-hour cadence resolves storm phases (quiet → main → recovery), not individual substorms.
Moran's I: inverse-distance haversine weights, validated against PySAL2 to machine precision.
Gjerloev, J. W. (2012). The SuperMAG data processing technique. Journal of Geophysical Research, 117, A09213. SuperMAG provides baseline-subtracted ground magnetometer data from a global network of stations with consistent processing. ↑
Rey, S. J. & Anselin, L. (2010). PySAL: A Python Library of Spatial Analytical Methods. In Handbook of Applied Spatial Analysis. Springer. PySAL's spatial autocorrelation module provides the reference implementation for Moran's I validation. ↑