# Letter | Steer clear of “collision avoidance”

In the story “Amazon gets key FCC approval for more than 3,000 LEO broadband satellites” (Feb. 8), the term “collision avoidance” is misused.

The implications of the term are that something the satellite operator did made any difference. Collision probability estimate is statistical. The threshold of 1 in 10,000 means that it is 9,999 times more probable that there would be no collision. If there were no collision, no one can prove that any maneuver made a difference.

We cannot “avoid” a collision. We can only diminish the probability of collision. Given the small threshold of probability of collision, we cannot bring it down below that. Only if the probability is about 1 in 10, can we reduce it much. It will always be much more likely that there would (not might) be no collision, and if there were no collision it does not prove that anything we did made a difference.

Orbit parameters are very uncertain, the further we look into the future the uncertainty grows rapidly. Calculation of collision probability is even more uncertain. That is why there are hundreds of thousands of conjunction data messages (CDM), while there have been no collisions.

The correct statement is that we can reduce the probability of collision. We cannot avoid a collision, and we should not claim that any maneuver “avoided” a collision. It is physically and mathematically impossible.

Therefore, Kuiper reporting regularly the number of maneuvers is meaningless. Reserving propellant for maneuvers has no value. It just reduces satellite life with no measurable effect.

Since orbit trajectories grow more and more uncertain, we cannot pin down the time or location of the “estimated” (not predicted) collision. Even worse, we propagate orbits in increments of hours, rarely as small as minutes because it would take a lot of computers to do that. For satellites with dimensions of even tens of meters, the collision event lasts only milliseconds. Propagation even in minutes cannot resolve such small intervals. It might be correct to give uncertainty bounds, which are much larger than the satellites or keepout distance.

This is not good news, but it is absolutely correct. To state otherwise inadvertently is misleading. To do so intentionally in order to influence perception of performance or resource allocation is more than tendentious. It is dishonest.

Please use “estimates” not predictions. Please never state that collisions were avoided or prevented. That is not possible, but the public thinks so because of perpetual misstatement.

With Respect,

Dr. David Finkleman, PhD
Col, USAF (Ret)