The Untracked Million
Every catalogue of space junk you have ever seen counts about forty thousand objects. That catalogue is the small part. The pieces big enough to end a satellite but too small for any radar to see outnumber it thirty to one — and nobody knows where a single one of them is.
ESA's statistical models put roughly 1.2 million objects between 1 and 10 centimetres in Earth orbit. At orbital closing speeds, anything in that band can destroy a spacecraft. Essentially none of them are individually tracked.
ESA Space Environment Report, 2025
What we can see, and what we can only estimate
Radar and optical surveillance networks catalogue objects down to roughly 10 cm in low orbit. Below that, the population is inferred from models, impact craters returned from spacecraft, and a handful of specialist radars. The numbers explode as the objects shrink.
Speed is what turns a fleck into a bullet
A satellite at 550 km travels about 7.6 km/s. Two objects in crossing orbits meet at a typical closing speed near 10 km/s — some thirty times faster than a rifle round. Kinetic energy scales with the cube of the diameter, so each step up the size ladder multiplies the damage a thousandfold.
Three hundred thousand dodges a year
SpaceX reports every automated collision-avoidance manoeuvre its Starlink satellites make to the US Federal Communications Commission, twice a year. Those filings are the clearest public measurement of how crowded low orbit has become — and the curve has bent almost vertical.
Up roughly 50% on the ~200,000 logged in 2024. Over the twelve months to mid-2026, more than 355,000.
Spread across the constellation, each Starlink now performs an avoidance burn on something close to a weekly cadence.
The crewed station has moved 41 times to dodge a tracked object — most recently in April 2025, for a fragment of a 2005 Long March upper stage.
Two events in three years made the modern debris belt
Most of the trackable junk in low orbit was not shed slowly. It arrived in single moments: a missile test in 2007, an accidental collision in 2009. Decades later the fragments are still up there, because at those altitudes the atmosphere is too thin to pull them down quickly.
Two-thirds of everything working up there is one company's
The shell of space between about 500 and 600 km has gone from near-empty to the busiest volume humans have ever operated in — in six years. Debris density in that band now approaches the density of active satellites.
A record. The previous high was 2,903, set in 2023. Nothing in the history of spaceflight resembles this rate.
Up 2,229 tonnes in a single year — roughly the mass of four fully loaded 747s added above our heads, every twelve months.
ESA's assessment of how much more dangerous low Earth orbit has become in two years. Its model projects debris density at key altitudes doubling by 2030.
The cleanup has a downside nobody costed
The fix for debris is to deorbit satellites fast, and it is working: more than three objects now re-enter every day, and controlled reentries outnumbered uncontrolled ones for the first time in 2024. But every satellite that burns up leaves its metal in the upper atmosphere.
Intact satellites and rocket bodies falling back, on average. 90% of LEO rocket bodies now meet the 25-year disposal rule; 80% meet ESA's stricter 5-year standard.
Aluminium oxide deposited in the mesosphere by re-entering satellites in one year — an eightfold rise since 2016, and 29.5% above natural background levels.
Alumina nanoparticles generated by a typical satellite's demise. They can take up to 30 years to drift down to the ozone layer, where they catalyse chlorine chemistry.
Donald Kessler and Burton Cour-Palais described the trap in 1978: collisions make fragments, fragments make collisions, and past a certain density the belt grows on its own. ESA's 2025 assessment puts it plainly — even if every launch stopped tomorrow, the debris population would keep rising, because fragmentations are outpacing what the atmosphere drags down.
Orbit is a commons with no sweeper
Low Earth orbit carries the weather forecasts, the navigation fixes, the crop monitoring and the disaster imagery that everything else runs on. It is also the only shell through which crewed spacecraft pass. The million pieces we cannot see do not make it unusable — they make it expensive, fragile, and dependent on everybody behaving.
- Deorbit deadlines are the leverThe shift from a 25-year to a 5-year disposal rule is the single biggest change anyone has made to the forecast. Compliance is already 80% for LEO rocket bodies; getting it to 100% matters more than any removal mission.
- No more deliberate breakupsA single ASAT test in 2007 added more catalogued debris than a decade of accidents. Over 150 nations have now backed a moratorium on destructive direct-ascent ASAT testing; it costs nothing and prevents the worst case.
- Fund the telescope, not just the rocketWe cannot avoid what we cannot see. The gap between 54,000 large objects believed to exist and 40,000 actually catalogued is a tracking problem, and tracking is cheap compared to losing a constellation.
- Watch the atmosphere tooSolving debris by burning everything up trades an orbital problem for a stratospheric one. The alumina measurements are only two years old; the honest position is that we don't yet know the ceiling.
Sources & notes
- ESA Space Environment Report 2025, ESA Space Debris Office — object counts (40,000 tracked, ~11,000 active payloads), size-class estimates (54,000 >10 cm; 1.2 million 1–10 cm; ~130 million 1 mm–1 cm, populations to Aug 2024), 13,579 t mass in orbit (+2,229 t year on year), >3 reentries/day, 90%/80% disposal compliance, 3,000+ catalogued fragments from 2024 fragmentations, ~10.5 accidental fragmentations per year, and the statement that debris would keep growing with zero further launches.
- ESA Space Debris Office / DISCOS statistics, 2026 update — LEO collision risk up ~20% since 2024; modelled doubling of debris density at key LEO altitudes by 2030; 550 km identified as the critical band. Modelled projection, not a measurement.
- J.-C. Liou, NASA Orbital Debris Program Office (2015), “An Analysis of the FY-1C, Iridium 33, and Cosmos 2251 Fragments” — 3,549 fragments from Fengyun-1C; 5,579 catalogued across all three events; ~5,000 still in orbit in Jan 2013. The Iridium 33 × Cosmos 2251 figure shown (~2,030) is derived by subtraction and is approximate.
- SpaceX semi-annual constellation reports to the US Federal Communications Commission, 2021–2026 — six-month manoeuvre counts (2,219 / 3,333 / 6,873 / 13,612 / 25,299 / ~49,300 / 144,404 / 148,696 / 207,152), ~300,000 manoeuvres in 2025 vs ~200,000 in 2024, 355,000+ over the twelve months to mid-2026, and the 3×10⁻⁷ manoeuvre threshold. Filings reported via Space.com.
- Jonathan McDowell, Jonathan's Space Report (planet4589.org), 2026 — 16,019 active payloads (June 2026); 11,102 Starlink satellites in orbit, 11,087 working (27 Aug 2026).
- NASA / International Space Station Program — 41 debris-avoidance manoeuvres since 1999, most recently 30 April 2025 (Progress 91 burn, 3 min 33 s, against a Long March 2D upper-stage fragment); ISS shielding effective to roughly 1 cm.
- Ferreira, Schulz, Plane et al., Geophysical Research Letters, 2024 — “Potential Ozone Depletion From Satellite Demise During Atmospheric Reentry in the Era of Mega-Constellations”: ~30 kg of alumina nanoparticles per 250 kg satellite; 17 t of aluminium oxide deposited in 2022; eightfold increase 2016–2022; +29.5% over natural levels; up to 30 years for particles to reach stratospheric altitudes.
- Kessler, D. J. & Cour-Palais, B. G., Journal of Geophysical Research 83, 2637–2646 (1978) — “Collision frequency of artificial satellites: The creation of a debris belt.”
- Our World in Data / UNOOSA Online Index of Objects Launched into Outer Space — 4,510 objects launched in 2025; previous record 2,903 in 2023.
ILLUSTRATIVE / CALCULATED FIGURES, FLAGGED: the kinetic-energy bars in section 02 are computed by us (½mv², solid aluminium sphere at ρ=2.70 g/cm³ and a 10 km/s closing speed) and are not agency-published values; the everyday comparisons beside them (baseball, car, TNT) are our own equivalences. The orbital-shell diagram in section 05 is schematic and not to scale. All sub-10 cm debris counts are modelled estimates — by definition, objects that small cannot be individually observed.