Reading the scene
green = harder spectrum → emission ~110 km
red = softer spectrum → higher, redder emission — cell heights anchored to the photo-MEASURED border altitudes for each epoch (110–360 km)
ray tops turn blue-violet where they rise into sunlight above Earth's shadow (N₂⁺ resonance — the Gannon signature)
SAR arc = smooth thermal red glow @ ~400 km, detached equatorward of the oval — drawn only over the longitude sectors with SAR-classified reports; night side only
red band low on the equatorward horizon = 630 nm red aloft (200–400 km) from aurora 1,150–2,200 km away — its green lies below your horizon; the ordering is distance, not altitude
equatorward bnd. poleward bnd.
pink ring = AMPERE field-aligned-current oval — CONTINUOUS 10-min measurement (Iridium constellation), alive between satellite passes
Pins: camera SAR SAR+visible unclassified
cyan markers = 486 star-calibrated cameras with recovered capture times (bright within ±30 min of their moment; tick shows pointing)
emerald arcs = photo-MEASURED auroral border/top heights (ensemble fits, e.g. border 360 km @ maglat 51.7° over N.America 05:30 UT)
a faint flat airglow floor is always present even with no aurora — the 557.7 nm nightglow, ~250 R (Leinert et al. 1998, Table 13); it is a real sky glow, not part of the auroral model
amber = light pollution: city lights on the night side (globe — NASA Black Marble) and, at your stand, modeled skyglow + horizon light-domes with a Bortle-class estimate — Lorenz World Atlas 2025 (Cinzano/Falchi propagation method on EOG/VIIRS annual lights, ~0.9 km grid); Bortle shown is estimated from modeled zenith brightness — Bortle proper is a whole-sky visual scale
pale drape = actual VIIRS night photo (aurora + city lights), near its overpass time
animated W-Canada patch = TREx RGB + THEMIS ground cameras: each circle is one camera's field of view, projected at an assumed 110 km (product limitation — red aurora above 110 km smears toward footprint edges); 5-min steps, rims feathered by view-elevation confidence
Measured vs. derived: flux, boundaries & timing are SSUSI measurements (per ~101-min orbit, no interpolation between passes). The
Modeled aurora layer is the fused normalization of every source into the sun-fixed magnetic frame: oval
position from AMPERE's measured 10-min field-aligned-current boundaries;
brightness/energy fused from DMSP/SSUSI FUV imaging and POES+MetOp TED electron precipitation (5 satellites in different local-time planes, cross-calibrated to FUV-equivalent flux — measured median ratio ≈8: TED's 50 eV–20 keV band misses the storm-time hard tail SSUSI sees), carried between samples in the sun-fixed frame with AE-indexed substorm scaling. Archive reconstructions are
presented on a 240×32 display grid bicubically interpolated from the 120×16 fusion — presentation smoothing only; the measured information content is unchanged. In Source mode, brightness dims with per-cell
confidence — how directly that sector & moment was sampled; the Simulation renders the best estimate at near-full brightness and reports confidence in the HUD instead. F18's pre-midnight blind sector and between-pass gaps are constrained by POES crossings where they exist. Dayside precipitation is real (imagers measure it in sunlight) but invisible to the eye — so all auroral layers fade with ground daylight using the same sun position as the terminator: the model fades out fully, measured snapshots dim to a ghost. Where the photo ensemble measured the visible red border & column tops (green lines), the reconstruction extends soft red emission equatorward to that measured border and lifts red-column tops to the measured heights. Sub-boundary soft red renders as SAR-regime diffuse glow — smooth, structureless, pure 630 nm (Kozyra+ 1997) — rather than curtains. In Simulation mode, fine ray structure inside the curtains is
procedural — real auroral rays are 0.1–10 km, below every instrument's resolution here — while position, brightness, extent and heights stay measured. The procedural structure's
spacing is fitted to our own corpus: 19 star-calibrated photographs measure a median fold-strand spacing of 15 km (the renderer draws 17 km) — a superstorm-corpus fit (May/Oct 2024); quiet-sky spacing is an extrapolation, as no published population study of fold spacing vs. activity exists; ray widths follow published distributions (Maggs & Davis 1968 → Yang+ 2024), which our photo resolution cannot yet independently verify. The fine-structure
spectral jitter amplitude self-calibrates per storm from the reconstruction's own measured energy texture (the same residual statistic that already calibrates the per-curtain spectral spread), normalized to the photo-validated 2024 flagship as the fixed point — storms whose measured spectra are genuinely smoother render calmer color texture; because fit noise can only inflate the measured roughness, downward calibration runs free while upward is capped at +15% (?ejit=0 pins the previous fixed amplitude). Column tops vary per cell: the photo-measured epoch envelope scaled by each cell's own measured flux & energy (Rees range–energy curve), so bright soft cells reach the measured maximum while weaker cells stand shorter. Where a column top rises above Earth's shadow (same sun geometry as the terminator), it turns
N₂⁺ blue-violet — deep-night sectors self-gate because the shadow there sits above every measured top. In ground view, cells near the observer resolve into
3D field-aligned columns at the dipole inclination, so looking straight up produces the true
corona convergence at the magnetic zenith — column placement, heights and brightness are the measured field; only sub-cell ray multiplicity is procedural.
Substorm flags (amber ▼ on the timeline, badge while active) are detected from the AE index: onset = ≥150 nT rise within 20 min from a non-rising baseline, expansion until AE stops making new highs. Arc
dynamics follow published morphology: omega-band brightness waves ride the poleward edge eastward at ~0.9 km/s during substorm-active hours (Opgenoorth+ 1983), giant-undulation waves ride the equatorward diffuse edge westward when SYM-H drops below −150 nT (Lui+ 1982, K-H of the SAID shear; the −150 nT gate is this engine's threshold for the deep-storm case-study regime — no population statistic exists), and the procedural fine structure drifts along the arc (curl/fold scales, Hallinan 1976; Partamies+ 2001) — all brightness-led, mean-neutral, with any displacement bounded inside one grid cell (sub-resolution). At the substorm surge head the arc
winds into a spiral — the westward-traveling surge is itself a spiral (Davis & Hallinan 1976: the 20–1300 km class) — with winding that rides precipitation intensity and a rotation sense always counterclockwise viewed along the field line in both hemispheres (Partamies+ 2001, 216-event interhemispheric study); the winding is a photon-conserving rotation of the drawn structure — flux is redistributed, never created. In instantaneous view the fold strands additionally carry
rolling curls — Kelvin-Helmholtz roll-ups with a grow–wrap–decay lifecycle of seconds (Hallinan & Davis 1970; Trondsen & Cogger 1998), wound opposite to spirals (Hallinan, Davis & Webster 1972); long exposures smear them, so the view keeps the time-averaged fold texture. Beyond the surge,
spiral streets ride every arc — trains of 25–75 km vortices at 125–175 km wavelength drifting with convection, most frequent in magnetically quiet postmidnight skies (Partamies+ 2001) — so the oval carries distributed swirl at all activity levels, in both hemispheres. The Classic simulation warps its curtain walls with the same surge-spiral rotation (one physical model, two renderers); spiral streets and curls stay Ultra-only because 25–75 km and 1–5 km forms sit below the Classic walls' ~140 km segment resolution, and Source mode shows measured snapshots — no synthetic swirl is ever drawn over measurements. The sheet itself carries
drapery meander: two traveling sinuous modes at ~150 km and ~450 km wavelength (the middle of the classic curl→fold→spiral rotational hierarchy) weave the band's centerline up to ~1° of latitude — drawn geometry at sub-grid amplitude, never a claim about measured placement — and the small vortices ride the crests of that wave, winding up as its envelope grows: Kelvin-Helmholtz roll-up as observed, not decoration. Omega-band tongues protrude poleward in step with their brightness wave (Opgenoorth+ 1983). The Classic renderer carries the 450 km mode (its wall segments cannot resolve 150 km). Kill switch ?drape=0. The whole dynamic hierarchy is
driven by the measured input data: a shear-drive index computed from the measured solar-wind coupling (Newell ε, 45-minute mean — arc deformation is the magnetosphere–ionosphere feedback instability, switched on and strengthened by convection drive, Watanabe+ 2016) and, where a density measurement is unambiguous (live solar wind, the Storm Lab dial, and the measured replays — whose solar-wind density is carried from the same NASA OMNI record the reconstructions were built from), a weak density factor (K-H vortex regimes depend on the density jump across the shear layer, Faganello+ 2008 — a model-derived regime scaling, not a measured auroral law) scales the weave amplitude, wave speed and wavelength, vortex winding and occurrence, and curl activity; the surge-spiral core grows with the measured onset magnitude (Partamies 2004). At nominal driving the index is exactly 1 and the calibrated look is unchanged — the data moves the dance around that reference, and the 1.1° displacement clamp is never exceeded regardless of drive. The drift frame itself follows the ionosphere's
two-cell convection with the Harang discontinuity: in the 19–02 MLT sector the flow reversal slants across the band in latitude (westward poleward, eastward equatorward — the midnight velocity shear zone, Koskinen & Pulkkinen 1995), so structure on either side of the shear is carried apart and the eddies concentrate on the shear line itself — where substorm onsets are in fact observed to cluster (Weygand+ 2008; pre-onset forms track this flow pattern, Nishimura+ 2010). The Harang region is itself
activity-dependent: as the measured convection drive strengthens the whole reversal window rotates duskward (up to ~1.3 MLT-h; Gkioulidou+ 2009 modeling and Zou+ 2009 SuperDARN/ISR observations of the reversal's equatorward-and-duskward migration), and in the ~50 minutes before a substorm onset the shear layer visibly sharpens (Zou+ 2009), riding the same pre-onset growth signal the field-line layer uses; at nominal drive the geometry is exactly the static configuration (?har=0 pins it). Kill switch ?flow=0. In lulls and substorm recovery the equatorward half of the oval hands over to
pulsating aurora — log-normal 30–200 km diffuse patches blinking on 2–20 s cycles at measured duty (Partamies+ 2017 400-event survey; chorus-decay timescales, Troyer+ 2024), drawn ~8 km lower, mean-preserving so long exposures hold the steady veil real cameras record; patches
drift with the same convection frame — eastward after midnight, westward before — at the measured E×B speeds (Yang+ 2015, 2017: patch motion matches SuperDARN convection), and a weaker amorphous-class population extends into the 21–24 MLT sector equatorward of an active breakup (Grono & Donovan 2018/2020 classification & occurrence surveys). Kill switch ?pa2=0. One macro structure is deliberately
not drawn: during sustained northward IMF the live panel names the possibility of a
transpolar arc (a closed-flux arc crossing the polar cap — Milan+ 2005, confirmed by Fear+ 2014 — seen in only ~10–16% of favorable intervals, Kullen+ 2002) and links real near-real-time polar imagery instead of rendering one: transpolar-arc occurrence at any given moment is a statistical draw, not a measurement, and this simulation draws only measured or corpus-anchored structure. (The same advisory names the
horse-collar configuration — dawn/dusk arcs closing poleward across the cap under sustained northward IMF, ~8 events/month, no seasonal or Bx dependence; Bower+ 2022.) During flagged expansions the simulation brightens a bulge using the published statistical template (Gjerloev+ 2007) with westward/eastward expansion speeds of 8.8/5.3 km/s (Ogasawara+ 2011; Craven+ 1989) — timing & amplitude come from our measured AE, onset location from the model's own brightest near-midnight sector. When an event's later intensifications fire, the synthetic builders can ignite a
secondary expansion front — a discrete curtain drawn poleward of the main band, the stepwise poleward leap of tailward reconnection retreat (THEMIS multi-onset observations) that persists into recovery as the double-oval poleward arc system (Elphinstone+ 1995); its flux is diverted from the owning intensification's own budget — redistributed, never created — and measured storms never draw it (they show what the instruments saw). Kill switch ?front2=0. In the synthetic builders the substorm
schedule itself is generated by a loading–unloading energy balance: a tail-energy integrator accumulates the scene's own measured/modeled coupling (Newell ε) and an onset fires when the stored energy crosses a drawn threshold, consuming it — the minimal substorm model (Freeman & Morley 2004; Klimas+ 1992), which reproduces the observed ~2.7 h inter-substorm waiting time under steady driving (Borovsky & Yakymenko 2017); release magnitude rides the stored energy, so quiet nights genuinely cannot afford storm-cadence reels. Measured storms never use this path (their onsets are ground-measured), and on tonight's resimulation measured onsets unload the integrator at their detected times. Kill switch ?subphys=0. Where the scene carries
measured IMF By (the resimulation's measured half; archived nights), each drawn onset's sector additionally shifts duskward for By>0 — the interhemispheric displacement organized by By in conjugate imaging (Østgaard+ 2004), applied at the observed scale (~0.12 h/nT, saturating ±1.2 MLT-h) to the Frey+ 2004 base statistics. The shift is the northern-hemisphere sense; the engine carries one sector per onset for both hemispheres, so the south inherits it (the real southern shift is opposite in sign — a disclosed single-sector limitation). Scenes without measured By (the forecast's Enlil driving, the Storm Lab) keep the pure statistics. Kill switch ?bysec=0. The 24-node boundary rings are drawn through a periodic cubic passing exactly through the same fitted nodes — piecewise-linear chords polygonized the oval from orbit, a node-pitch artifact, not a measured shape (?ringcr=0 restores the polyline) — and the instantaneous-view brightness seethe collapses toward its ensemble mean as a pixel's footprint spans many independent curtains, so the top-down macro view shows the calm ensemble an orbital camera records while ground-view curtains keep the full 10-second seethe (?flick=0 restores the previous behavior). Emission colors follow the Rees & Luckey (1974) energy–ratio relation; peak altitudes follow the Fang et al. (2008) Maxwellian energy-deposition parameterization (P
ij verified against three independent sources), epoch-scaled by our photo-measured heights (storm-time thermospheric expansion, cf. Kataoka+ 2024, who measured Gannon red to ~1000 km — consistent with our independent 790–870 km ensemble tops); ray structure fades with altitude per the 630 nm ~110 s radiative lifetime; the sunlit-blue gate (−24° solar elevation) is derived from the render's own shadow geometry — storm-time blue emission measured at 400–900 km altitude (Nanjo & Shiokawa 2024) is consistent with these sunlit tops. The
naked-eye view applies scotopic/mesopic vision to the simulation: rod sensitivity peaks at 507 nm and vanishes toward 630 nm (Purkinje), so faint aurora renders dim gray-green and red appears only where bright enough to engage cone vision — matching the naked-eye vs photograph color reports collected by Grandin+ 2024. Default long-exposure shows what cameras record.
Photo-POV: click any calibrated camera dot to stand at its exact spot & time with the plate-solved frame outlined on the sky; gold dots blend the actual photograph over the render for direct validation. Frames are auto-leveled using the photo's own horizon (the level-camera assumption that also powers the star clock) — untick "level" to see the raw solved orientation at the assumed exposure time; the residual tilt there mostly measures clock error (parallactic rotation), not camera tilt. Emission altitudes: hard/green cells at ~110 km; soft/red cells lifted to the photo-ensemble MEASURED border heights per epoch (110–360 km across the storms). Tall-ray
placement follows the observed soft-precipitation flank (illustrative until photo triangulation replaces it) — but ray
altitudes (500–1100 km) and the red→blue sunlight transition use real shadow geometry per location & time. Photo triangulation has begun: two same-minute corpus photographs (Muszyna & Kłonów, Poland, May 11 2024, 122 km baseline) stereoscopically measure the auroral lower border at
99 ± 5 km, independently confirming the model's ~100 km base height (method in the project's research notes).
Live nowcast: the predicted
visible border blends the measured solar-wind coupling (Newell ε, 45-min mean) through relations fitted on our two photo-validated 2024 superstorms, with the quiet end anchored to the Carbary (2005) Kp-oval climatology and measured Kp outranking the storm-fitted estimate at low activity. The live
SAR arc is a model band: placed at the Carpenter & Anderson (1992) plasmapause (their max-24-h Kp form), intensity ramped on the measured ring current (SYM/Dst; Kozyra+ 1997 energetics), brightness classes ~0.1–6 kR and the ~400 km altitude per Martinis+ 2022, drifting with the Dst trend inside the measured −100…+200 m/s range (Takagi+ 2018); outside storms a very faint substorm-associated band can render (Gololobov+ 2023). Inbound-CME
arrival calls are CCMC analyst WSA–Enlil runs (DONKI); their typical arrival error is ±8 h (Wold+ 2018). SAR band: latitude fitted to SAR-classified reports in Chris Wicklund's sightings map, drawn only over reported longitude sectors; alt. ~400 km per FRIPON Europe obs. Sighting pins are event-integrated.
The Moon renders at its true ephemeris direction and phase for every scene's clock (truncated Meeus series, verified to <1′ of the reference example) — each historical storm shows its historically correct moon. Distance is compressed to 9 R⊕ and the disk drawn ~35% over-size for readability; moonlight lightens the night hemisphere in proportion to the illuminated fraction. The mag-lat grid uses the same centered/tilted-dipole transform as the border rings (adequate at auroral latitudes; the real IGRF field diverges most near the South Atlantic Anomaly).
Field lines are traced through the dipole plus the Tsyganenko (1989) empirical external field — tail current sheet, ring current and magnetopause current systems, parameterized by activity level — driven by each scene's own conditions: Kp from the scene's measured/modeled solar-wind coupling (the same ε→Kp relation the builders use, measured Kp outranking it in live mode) and dipole tilt from the scene clock, re-traced as conditions change. The tail visibly stretches as energy loads and the dayside compresses; lines the model classifies as open (polar cap / tail lobes) render dim violet. T89 is a statistical average configuration for a given activity level — a published model, not an instantaneous measurement (?t89=0 restores the static dipole L-shells). During flagged substorms the configuration additionally follows the
loading–unloading cycle: the tail stretches through the growth phase and relaxes dipolar at onset, keyed to the scene's own onset record (ground-measured onsets on measured storms and live; the energy-balance schedule on synthetic scenes) — implemented strictly as interpolation between the model's published activity levels, a
model-derived illustration of dipolarization, since T89 itself carries no substorm time dependence (?dip=0 pins the static level).
In memory of Jennifer Gannon. The May 2024 superstorm carries her name — this observatory began as its reconstruction.
Data: NASA SPDF (SSUSI, OMNI) · NASA GIBS (VIIRS DNB · Blue Marble globe) · AMPERE Science Data Center (JHU/APL, NSF; PI B. Anderson) · UCalgary AuroraX (TREx RGB + THEMIS ASI grid mosaics) · sightings: Chris Wicklund's maps (Wicklund et al. 2024, OSF e87cs) · ground view: map © OpenStreetMap contributors © CARTO, terrain Mapzen/AWS Terrain Tiles.