The increasing quantity of debris orbiting the Earth is causing concern for space agencies. At the average collision velocity of 10 km s−1, even projectiles 1 cm across can damage satellites. Impacts have a significant chance of occurring, especially for large structures that remain in orbit for a long time1. Particularly at risk are satellites at altitudes of about 800 and 1,400 km, where there is already a high density of orbiting bodies and atmospheric drag is not effective in removing small fragments. We find that the dynamical architecture of satellite constellations, several of which are to be launched over the next decade, makes them particularly vulnerable to the consequences of an impact, which may set up a kind of chain reaction, triggering more collisions.

There are currently about 104 objects larger than 10-20 cm across orbiting the Earth, with 105 objects more than 1 cm across, and 107 bodies exceeding 1 mm in diameter. These have a total mass of about 3×106 kg and a cross-section of 4×104 m2 (refs 2,3), and the average flux (the expected impact rate per projectile and per unit target cross-section) is about 3×10−10 m−2 yr−1. In the long term, debris from collisions with these objects may irreversibly pollute portions of near-Earth space4,5.

Satellite constellations, consisting of tens to hundreds of satellites orbiting at about the same altitude, will be launched into orbit over the next ten years, mainly for global telecommunication purposes6. One of these, the Iridium system, is already operational. It consists of 66 satellites (plus six spares) orbiting in six different orbital planes at an altitude of about 780 km and with an inclination of 86.4 degrees. The wet mass of each satellite is 667 kg. Although some measures have been taken to minimize the risk of impact, the possibility of disruptive collisions cannot be ruled out, as the selected altitude corresponds to a peak of the expected debris flux. Assuming a catastrophic threshold for the impact specific energy of 4.5×103 J kg−1, as inferred from laboratory experiments on spacecraft mock-ups7, the flux of potentially shattering fragments at 800 km is about 1015 m12 yr−1. For a constellation cross-section of about 103 m2, the risk of a catastrophic impact is about 10% per decade.

The architecture of such a constellation will make a break-up event particularly dangerous, owing to the spreading of the resulting fragment swarms, on a timescale intermediate between those analysed in previous studies (several hours8 and several decades9). This is particularly true when differential precession of the orbits leads the fragments to encounter satellites revolving around the Earth in the opposite direction (Fig. 1); this makes head-on collisions possible, with higher impact speeds and greater collision probability.

Figure 1: The orientation of the six orbital planes of an Iridium—like constellation, as seen from the celestial north pole.
figure 1

The revolution of the satellites is indicated by the arrows. Note the ‘counter—rotating’ planes 1 and 6. If a satellite orbiting in either of these planes is disrupted, its fragments will pose an impact hazard much higher than average to the satellites in the other plane.

Results of a simulation of an Iridium satellite being broken up by a 1-kg projectile at a relative speed of 10 km s−1 are shown in Fig. 2. We have modelled the resulting fragment swarm (8,000 bodies heavier than 1 g) under Earth's oblateness and air drag perturbations, and computed as a function of time the cumulative probability of impact with another constellation satellite moving in the opposite direction.

Figure 2: Flux of fragments over time produced by the simulated impact break-up of an Iridium satellite orbiting in plane number 6 of <.
figure 2

Fig. 1. Panels show, from top to bottom, the fluxes at decreasing kinetic energy, E : 2E >109 J; 108<2E <109 J; 107<2E <108 J; 106<2E <107 J; 105<2E <106 J; 104<2E <105 J; and 103<2E <104 J. The target satellite is located in plane number 1 (Fig. 1). Horizontal dotted lines represent background flux in the same energy ranges, calculated from the overall space debris population, according to ref. 2.

For an impact energy between 107 and 108 J, corresponding to such projectiles, the collision probability stays higher than the background level from the general orbiting population for several years. This means that, after the initial break-up, the probability that a second one will follow within five years is about 10%. This may then trigger a chain-reaction effect with a characteristic timescale of about a century, much less than the current estimates with the general debris population (300 to 500 yr)10.