Space debris: what is up there, and what takes it away

Every object in this article is one the world's tracking networks can see individually and follow from pass to pass. In low orbit that means roughly grapefruit size and up; higher up, larger still.

The catalog splits into three kinds of thing. Payloads are the satellites themselves, working or dead. Rocket bodies are the spent upper stages that carried them up and stayed in orbit afterwards. Debris is everything else: fragments from breakups, plus the covers, clamps and adapters released during normal operations.

The first surprise in the numbers is how much of the debris is already gone. Of the 34,567 fragments catalogued since 1957, 22,231 have reentered and burned up. They were never removed by anyone. They simply ran into the top of the atmosphere, over and over, until they came down.

Orbital debris is not evenly produced. It arrives in bursts. A single satellite destroyed at the wrong altitude can add more trackable objects in one afternoon than a decade of routine launches, and the biggest clouds are still individually recognisable in the catalog decades later.

The very first one set the pattern. On 29 June 1961, the upper stage that launched the Transit 4A navigation satellite exploded 77 minutes after liftoff, and 310 tracked pieces more than tripled the size of the entire catalog overnight. 58% of that debris is still in orbit today, 65 years later, because it happened high.

The lesson took a while to land. In 1986, the Ariane 1 third stage that had delivered SPOT-1 nine months earlier burst from leftover propellant, scattering 496 fragments through the sun-synchronous highway. That event is why passivation, deliberately venting every tank and discharging every battery once a stage has finished its job, became standard practice worldwide.

Debris does not sit where the traffic is. Plot everything in orbit by altitude and two distinct populations appear, and they barely overlap.

Intact satellites cluster low, peaking in the 400 to 500 kilometre band where the broadband constellations fly. Debris peaks two hundred kilometres higher, at 800 to 900 kilometres. The busiest shell in space is not the dirtiest one.

The atmosphere does not stop at a line. It thins out gradually, and even at hundreds of kilometres up there is enough gas left to matter. A satellite at 400 kilometres is flying through something like a hundred-trillionth of sea-level air, but at 7.7 kilometres per second, encountered continuously for years, that whisper is a brake.

Every pass steals a little energy. The orbit shrinks, which makes the satellite go faster and dip into slightly thicker air, which steals energy faster. The process accelerates until the object meets real atmosphere and breaks up in a few seconds of heat. Almost everything burns; the rest lands, usually in an ocean.

How long that takes depends on altitude far more than on anything else, and the effect is dramatic. Below is the same measurement for five clouds: the share of each still in orbit, counted from the day it was created.

The chart above deliberately lines the clouds up at their moment of creation, which is the right way to see the altitude effect and the wrong way to see anything else. Put the same curves back on real dates and a second pattern appears that the first view hides.

The atmosphere is not a fixed thing. The Sun runs an eleven-year activity cycle, and near its peak the extra ultraviolet and X-ray output heats the upper atmosphere and puffs it outward. The air a satellite meets at 800 kilometres near solar maximum can be many times denser than the same altitude near solar minimum. Every cloud in orbit feels that at the same time, which is exactly the shared bending in the chart above.

Isolate one cloud at one altitude, count how many of its fragments reenter each calendar year, and the rhythm draws itself with no analysis at all.

None of this is left to luck any more. Four habits do most of the work, and all four are now normal engineering rather than aspiration.

Passivate everything. A spent stage still holds propellant, pressurised gas and charged batteries: exactly the ingredients of the explosions that made several of the clouds above. Venting and discharging them at end of mission removes the energy that would otherwise turn one object into a thousand.

Come down quickly. The long-standing guidance was to clear low orbit within 25 years of end of mission. In 2022 the US regulator cut that to five years for new licences, and Europe now expects the same. Operators increasingly design for it directly: fly low enough that the atmosphere alone finishes the job even if the spacecraft dies.

Move out of the way. Conjunction warnings go out when two catalogued objects are predicted to pass close, and satellites with propulsion manoeuvre. For a large constellation this is thousands of automated avoidance burns a year, handled without human intervention.

Go and fetch the big ones. Removal has moved from concept to flight hardware: magnetic capture of a client satellite demonstrated in orbit, a close-up rendezvous and inspection of a real discarded rocket stage, and a contracted mission to remove a piece of hardware outright. Each flight retrieves one large object, which is the right target: a massive derelict stage is not just debris, it is a future breakup waiting to happen.

In 1978 Donald Kessler described a scenario in which collisions between objects in orbit generate enough fragments to cause further collisions, and the debris population becomes self-sustaining. It is a real piece of physics and worth taking seriously.

What the catalog shows is a system with two competing rates. Fragments are added in bursts by breakups; they are removed continuously by the atmosphere, faster the lower they are. In the shells below about 600 kilometres, where nearly all the new traffic is going, the removal rate is fast enough that the population cannot run away: a failed satellite there reenters within years whether or not anyone intervenes.

The band that deserves the attention is the one this article keeps returning to, roughly 700 to 1,000 kilometres, where removal is slow and the historic clouds already sit. That is precisely why the current rules push new hardware lower, why passivation is mandatory, and why the first removal missions are aimed at the biggest derelicts up there.

The trend that matters most is the one that does not photograph well: things that did not happen. Stages that were vented instead of exploding. Satellites that were flown down instead of abandoned. Close approaches that ended in a small burn. Orbit is getting busier and, at the same time, better run.