Meteor Ablation: Why a Grain of Dust Makes a Streak

The streak is not the rock

Start with what a meteor cannot be. Take a millimetre grain glowing at its ablation temperature of about 2,200 K. Its surface area is around 3×10^-6 m², so as a blackbody it radiates roughly σT⁴A ≈ 4 W, with the Wien peak at 1.3 µm — deep in the infrared. From 100 km away that is invisible, and it is the wrong colour. Yet the same grain routinely produces a naked-eye meteor.

The resolution is that the emitting object is not the grain. It is a plasma column. Atoms boiled off the surface leave at thermal speeds of order 1 km/s but inherit the meteoroid's velocity, so each ablated iron or sodium atom ploughs into the surrounding air at 20-70 km/s. Every collision carries tens to hundreds of electronvolts, far more than the few eV needed to promote a bound electron. The metal atoms are collisionally excited, radiate, are excited again, and are eventually ionised; air molecules are excited and dissociated alongside them. The light is the sum of countless de-excitations, spread through a cylinder whose initial radius radar measurements put at roughly 0.3-1.5 m at 85-100 km, and which then spreads by ambipolar diffusion.

Because the body covers 10-40 km of sky during the 0.2-2 s that ablation lasts, that thin cylinder is painted across ten to thirty degrees of sky. The streak is a trail of air that briefly held a plasma, not a trajectory of a burning stone.

Free-molecular flow: molecular hammering, not friction

Above about 100 km the air is not a fluid as far as a sand grain is concerned. The molecular mean free path is around 2 m at 120 km, 0.14 m at 100 km and only 4 mm at 80 km. A millimetre meteoroid therefore has a Knudsen number far greater than one over the whole upper part of its flight: there is no bow shock, no boundary layer and nothing to rub against. Molecules arrive one at a time, strike the bare surface, and either stick, bounce or knock atoms loose.

The energy per hit is the whole story. A nitrogen molecule (28 u) at 40 km/s carries ½mv² ≈ 3.7×10^-17 J, about 230 eV; at 70 km/s it is over 700 eV. Surface binding energies are a few eV, so each impact can eject several atoms directly. This is sputtering, and it begins well before the body is hot — models by Rogers, Hill and Hawkes (2005) show it dominating above ~100 km and explaining why fast meteoroids start glowing so high. Leonid fireballs during the 1998-2002 storms were tracked from altitudes near 200 km.

As the density climbs, deposited energy outruns radiative cooling and the surface reaches the silicate melting and vaporisation range, ~1,800-2,500 K. Now thermal ablation takes over, and it is selective. Volatile sodium leaves the melt near 1,600-1,800 K, magnesium, silicon and iron follow, and refractory calcium, aluminium and titanium only at the highest temperatures. This differential ablation is why the sodium in a meteor spectrum peaks earlier and higher along the track than the calcium.

The ablation equations, with the numbers put through them

Classical single-body theory, in the form Ernst Öpik and later Zdeněk Ceplecha used, is two coupled equations. Drag: m dv/dt = −Γρ_av²S, with Γ ≈ 0.5-1 and S the cross-section. Mass loss: dm/dt = −(Λ/2ζ)ρ_av³S, where Λ is the heat-transfer coefficient and ζ the heat of ablation, about 6-8 MJ/kg for stone.

Dividing one by the other kills the atmosphere entirely and gives dm/m = σv dv, so m = m₀ exp[−σ(v₀² − v²)/2]. The single ablation coefficient σ = Λ/(2ζΓ) runs from ~0.005 s²/km² for irons and strong stones to 0.02-0.1 for fragile cometary material. Put σ = 0.014 and v₀ = 40 km/s in: slowing by a single km/s already costs about 40% of the mass. Small meteoroids do not decelerate then evaporate; they evaporate long before they decelerate.

Two more numbers explain everything else. First, kinetic energy per kilogram at 40 km/s is v²/2 = 800 MJ/kg, roughly 130 times the ablation enthalpy — survival is never an energy question, only a heat-transfer question. Second, only a sliver of that energy becomes light. The luminous efficiency τ is ~0.1-2%, rising steeply with speed, and the light curve follows I = −τ(v²/2) dm/dt. On the usual convention a zero-magnitude meteor radiates roughly 500 W, so about 30 mg of rock — a grain 2-3 mm across — makes a meteor as bright as Vega.

Survival follows from a third relation: a body loses its cosmic velocity once it has swept up an air column comparable to its own mass per unit area, m/S. A 1 kg stone has m/S ≈ 150 kg/m², met near 25-30 km altitude; a metre-scale body carries ~2,000 kg/m² and stays fast to ~15 km. Below the retardation point, at speeds under ~3 km/s, the glow stops and dark flight begins: several minutes of ordinary free fall at 50-100 m/s.

Colour is chemistry, and the atmosphere keeps the receipts

Meteor spectra are emission-line spectra, and Jiří Borovička's two-component model (1993) fits them with a main component near 4,500 K, which produces the neutral metal lines, plus a hotter ~10,000 K component that switches on above roughly 35 km/s and supplies ionised species. The orange-yellow that dominates most bright meteors is the sodium D doublet at 589.0 and 589.6 nm. Green comes from the Mg I triplet at 516-518 nm, violet from Ca II H and K at 393.4 and 396.8 nm, and a forest of Fe I multiplets crowds the blue and violet. Red is atmospheric: N₂ first-positive bands and the O I lines at 630 nm, with the forbidden green 557.7 nm line appearing in wakes and trains.

Because ablation is differential, spectra classify meteoroids. Borovička's scheme (2005) separates normal chondritic, iron, Fe-poor and, notably, sodium-free meteoroids — the last associated with orbits that pass close to the Sun, where solar heating had already baked the sodium out before Earth ever met them.

The atmosphere stores the evidence. Ablated metals do not vanish; they form permanent layers of Na, Fe, Mg⁺, Ca and K between about 80 and 105 km, peaking near 90 km, mapped nightly by resonance lidar and from orbit. The mesospheric Ca/Na ratio sits roughly 100 times below the chondritic value, exactly as differential ablation predicts. What recondenses becomes 1-2 nm meteoric smoke particles, which drift down and act as the nuclei of noctilucent clouds. The same sodium layer is what a 589 nm laser guide star is aimed at — adaptive optics on the world's largest telescopes runs on meteor exhaust.

How meteor ablation is actually measured

Triangulation from two or more stations turns a streak into a trajectory, a speed and a heliocentric orbit. Fred Whipple's Harvard Super-Schmidt cameras did it photographically in the early 1950s. On 7 April 1959 the Czechoslovak network caught the Příbram fireball and Ceplecha derived the first orbit ever tied to recovered stones — and, in a famous twist, Neuschwanstein fell on a nearly identical orbit on 6 April 2002 yet is an EL6 enstatite chondrite while Příbram is an H5, a compositional mismatch still argued over. The Prairie Network delivered Lost City in 1970, and the modern descendants are the European Fireball Network, the Desert Fireball Network, FRIPON, NASA's All-Sky Fireball Network, CAMS and the volunteer-built Global Meteor Network. Winchcombe (28 February 2021) was reconstructed and its CM2 fragments picked off a Gloucestershire driveway within twelve hours.

Radar sees what cameras cannot. High-power large-aperture systems such as Arecibo's 430 MHz dish (until its 2020 collapse), Jicamarca, the MU radar and EISCAT detect the head echo, the compact plasma ball travelling with the body, down to microgram masses. Transverse-incidence radars such as CMOR in Ontario see specular reflections off the trail: underdense trails, below ~10¹⁴ electrons per metre, ring and decay at a rate that measures ambipolar diffusion, so meteor radars double as mesospheric wind and temperature instruments; overdense trails reflect like metal cylinders and can persist for minutes.

Absolute calibration remains the weak link, which is why controlled re-entries of known mass and speed — Stardust's sample return capsule in 2006 and Hayabusa in 2010, both chased by airborne spectroscopy campaigns — are treated as artificial meteors for pinning down τ. For the largest events, US Government sensors feed the CNEOS fireball database, though the canonical Chelyabinsk numbers — 19.16 km/s entry, ~500 kt, peak brightness at 29.7 km — come from the combined video, infrasound and sensor analysis of Brown et al. (2013); the CNEOS entry itself lists 18.6 km/s and 440 kt.

Look-alikes, failure modes and open questions

The friction myth. Nothing rubs. For large bodies the heating is compression: gas piled up in the bow shock, radiating and convecting into the surface. For small ones it is bare molecular impact. Calling either friction predicts the wrong altitude, temperature and spectrum.

Vocabulary that gets mangled. Under the IAU's 2017 definitions a meteoroid is 30 µm to 1 m; the meteor is the light; a surviving stone is a meteorite. A meteor brighter than magnitude −4, about Venus at its brightest, is a fireball; a bolide is classically a fireball that detonates or shows a terminal flash, and a superbolide exceeds −17. Re-entering hardware is the commonest false alarm: it arrives at only ~7-8 km/s, tracks slowly and nearly horizontally for tens of seconds, and sheds a train of similarly coloured fragments.

The small survive best. Particles below ~50 µm have such a high area-to-mass ratio that they decelerate above 100 km and radiate their heat away before melting. That is why unmelted micrometeorites are recoverable from Antarctic ice while a gram-sized pebble is annihilated.

Still open. Faint meteors begin higher and brighter than single-body theory allows, which motivated the dustball model of Hawkes and Jones (1975) — loose aggregates releasing grains early. The luminous efficiency τ and the ionisation coefficient β are each uncertain by factors of a few, and that uncertainty propagates straight into Earth's dust budget: LDEF crater counts gave ~110 t/day, mesospheric metal chemistry favours ~40 t/day, and the published spread runs 5-300 t/day. The chemistry of persistent trains, thought to involve FeO* and NaO chemiluminescence fed by ozone, is only partly pinned down; the VLF explanation for electrophonic sounds proposed by Colin Keay in 1980 is still being tested; and claims that particular bolides, including CNEOS 2014-01-08, arrived on hyperbolic interstellar orbits remain contested on the grounds of unpublished velocity errors.

One case captures all of it. On 10 August 1972 a body several metres across skimmed to a minimum altitude near 57 km over Utah, Wyoming and Montana in broad daylight, ablated furiously for about 100 seconds, and climbed back out over Alberta. It lost mass without ever losing its heliocentric freedom.

What the atmosphere does to an incoming stony body, by size

Size / mass	Flow regime and heating	Where the light peaks	What reaches the ground
Under ~50 &micro;m (~10<sup>-7</sup> g)	Free-molecular; decelerates above 100 km and radiates heat away faster than it gains it	No visible meteor at all	Unmelted micrometeorites, collected from Antarctic ice and rooftops
~1 mm (~2 mg) grain	Free-molecular down to ~80 km; molecular impact and sputtering, then total vaporisation	~95-85 km; magnitude ~+4 at 20 km/s, ~+1 at 60 km/s	Nothing but nanometre meteoric smoke
~1 cm (~1 g) pebble	Free-molecular until ~80 km, where a bow shock finally forms	~80-70 km; magnitude ~-1 to -4	Nothing; ablation goes to completion
~1 m (~2 t)	Continuum flow, strong bow shock, fragmentation and flares	~40-25 km; magnitude ~-14	Kilograms of meteorites, after a dark flight of several minutes
~19 m (Chelyabinsk, 1.3&times;10<sup>7</sup> kg)	Airburst; ~500 kt released mostly as radiation and blast	29.7 km; absolute magnitude ~-28	~0.03% of the mass; the largest piece, 654 kg, from Lake Chebarkul
~50-60 m (Tunguska, 1908)	Airburst at 5-10 km; ~3-15 megatons	5-10 km	No confirmed macroscopic meteorites; 2,150 km&sup2; of flattened forest