What Causes the Northern Lights? Earth's Own Electrons

At about 04:04:30 Universal Time on 26 February 2008, a camera at Sanikiluaq, a Nunavut community on Flaherty Island out in Hudson Bay, recorded a brightening arc begin to surge poleward across its field of view. The instrument was one of the ground-based all-sky imagers built for NASA’s THEMIS mission, and Chang and Cheng later published the sequence in Earth, Planets and Space.

The energy behind that glow came from the Sun. The electrons that made it did not.

What causes the northern lights?

Electrons stored in Earth’s magnetosphere are accelerated down the planet’s magnetic field lines into the upper atmosphere, where they strike oxygen and nitrogen atoms and knock them into excited states. The atoms shed that energy as light. Solar wind supplies the power, but the glowing electrons come from Earth’s own magnetic tail.

That last clause is what fixes the timing. The store of electrons sits on the night side of the planet, which is why the sky performs on a schedule tied to your local midnight.

The solar wind’s job is energy transfer

NOAA’s Space Weather Prediction Center, which issues the US aurora forecast, puts it flatly in its own aurora tutorial: “It is technically not the solar wind electrons that create aurora.”

The handover depends on which way the Sun’s magnetic field happens to be pointing when it arrives. The solar wind carries an embedded Interplanetary Magnetic Field, and NOAA’s tutorial says that when that field is “aligned opposite Earth’s magnetic field (southward)”, “there is a transfer of more solar wind energy into the magnetosphere which accelerates more of the magnetospheric electrons down the magnetic field towards Earth.” A fast solar wind pointing the wrong way does far less.

“The auroral electrons that create the best aurora in the middle of the night actually come from the tail of the magnetosphere, downstream (away from the sun),” NOAA’s Space Weather Prediction Center states.

Earth keeps the fuel in its own tail

The magnetosphere is not shaped like a bubble. On the night side it is drawn out into a magnetotail, described by Borovsky and Valdivia in a 2018 review in Surveys in Geophysics as “a very long (100’s of RE), cylindrical volume of magnetic-field lines connected to the Earth” that works as “a reservoir of magnetic flux and energy.”

Inside it sits the plasma sheet, a hot, thin population of electrons. That same review states the point in one line: “Most aurora are produced by electron impact from the electron plasma sheet.”

The geometry explains the standard viewing advice. NOAA tells observers that “Best aurora is usually within an hour or two of midnight (between 10 PM and 2 AM local time)”, because at local midnight your patch of ground is turned to face straight down the tail.

The trigger is a snap 20 Earth radii out

Energy loads into the tail, then lets go all at once as a substorm. In 2008, Angelopoulos and colleagues published “Tail reconnection triggering substorm onset” in Science, using simultaneous measurements at several distances down the tail. Magnetic reconnection was observed at 20 Earth radii, around 127,000 km out on the paper’s own figure of 6,374 km per Earth radius, “at least 1.5 minutes before auroral intensification” and about three minutes before near-Earth current disruption. Their conclusion: “substorms are likely initiated by tail reconnection.” The disturbances behind intense auroral activity reach the ground as well: the magnetic observatory at Sitka has recorded erratic deflections of over 10 degrees in the horizontal field during them, enough to swing a compass needle.

The ordering is still argued over. A 2025 Nature Communications paper by Wu and colleagues calls the expansion phase “one of the most debated aspects of substorm dynamics”, and sets the near-Earth neutral line model, with reconnection “at distances of approximately 10-20 Earth Radii (Re) downtail in the plasma sheet”, against a current disruption model where near-Earth instabilities fire first and the disturbance works outward. Which event fires first is still an open question, and it is not the only ordering problem science has had to fight over.

Green is light an oxygen atom is forbidden to emit

When an electron slams into an oxygen atom high in the atmosphere, it can leave that atom in an excited state called 1S. NOAA: “The most common auroral color is a pale green color at a wavelength of 557.7 nm. This is the result of atomic oxygen having been excited to the 1S, or singlet S, state.”

Even that green often goes unseen. A faint display reads as grey or white to the naked eye, which NOAA attributes to night vision, “where the eye first picks out brightness without distinguishing color”, the same sort of edit that makes the vessels under your skin look like a colour they are not.

The 1S transition is a forbidden one. The team reporting the first detection of visible-wavelength aurora at Mars, writing in Science Advances in 2025, refer to “the forbidden atomic oxygen transitions at 557.7 and 630 nm.”

Forbidden is a statement about probability. The atom does make the jump, it is simply slow about it, and NOAA puts the 1S lifetime at “about 1 second”. Down where the air is dense, another particle reaches the atom inside that second and carries the energy off as motion instead, so no photon is ever released. The University of Alaska Fairbanks Geophysical Institute puts the bottom edge of the aurora at “about 60 miles (100 kilometers) above the surface of the Earth.” That floor marks the depth at which oxygen runs out of time.

The colours are an altitude map

Red comes from oxygen in a different excited state, 1D, which NOAA says has “a very long lifetime (>150 sec)”. An atom in that state needs air thin enough to be left undisturbed for over two minutes, which pushes red to the very top of a display. Molecular nitrogen emits promptly and needs no such quiet, so it paints the purplish lower border.

Colour	Emitter	Wavelength	Altitude band
Pale green	Atomic oxygen, 1S state	557.7 nm	120 to 400 km
Deep red	Atomic oxygen, 1D state	630 nm	Above 300 km
Purple lower border	Molecular nitrogen	Prompt emission	120 to 200 km

Altitude bands are NOAA’s; the 630 nm figure is the one used in the Mars aurora paper. Colour keeps turning out to be atomic bookkeeping rather than decoration, right down to the pigment on a farm building, which traces back to iron forged inside dying stars.

The direct-hit version is a simplification

The account you will find nearly everywhere runs like this: particles leave the Sun, cross the distance to Earth, get funnelled toward the poles, and hit our air. It is a compressed teaching version that hardened into a fact, and the compression deletes the storage step where all the interesting physics lives.

The honest complication is that the plasma in the magnetosphere does have solar ancestry. Borovsky and Valdivia list two sources for it: solar wind crossing into the magnetosphere at the magnetopause and cusps, and outflow from Earth’s own ionosphere. Material arrives from both. What it does not do is arrive and glow. It is captured, parked in the tail, and only later accelerated back down the field lines with enough energy to excite an atom.

Every display is stored charge coming home

Earth’s magnetic field banks solar wind energy in a tail hundreds of Earth radii long, builds it up slowly, then spends it in a substorm of roughly half an hour, and the light you see is the receipt.

The sky over Sanikiluaq that night was lit by particles that had been sitting on Earth’s own night side, waiting for a magnetic snap to send them back down. The Sun paid for the show. The timing, the storage and the green were all local. Strictly FYI.