3 ms·
I agree. Hard to believe those odds are credible even 24 hours out from the event at the velocity these objects were traveling. 1 in 23 billion are fantastical
by zaroth 6y ago
I agree. Hard to believe those odds are credible even 24 hours out from the event at the velocity these objects were traveling.
1 in 23 billion are fantastically long odds - roughly 8 sigma. What would the confidence intervals on direction, speed, and drag measurements have to be to put the possibility of a collision during an expected 70 meter flyby at relative velocity of 50,000 kph that far out?
How exactly predictable even is drag at that altitude on space junk which probably doesn’t have attitude control?
For example, I think LeoLabs 1 sigma confidence interval was +/- 18 meters, but they had calculated them crossing at 25 meters apart.
- colechristensen 6y agoYou basically just keep observing the orbiting object and add data to refine the orbital parameters. Orbital mechanics is very nonintuitive, there is very little drag at that altitude, and the speed doesn’t really have any influence on how precise you can be. With very few perturbations the object-in-motion-stays-in-motion becomes a powerful tool because you can use the fact that nothing changes to get crazy high precision by spreading your measurements over time.
- dmurray 6y agoFrom your numbers, you'd need to be about twice as good as LeoLabs. A one sigma CI of +/- 9 metres and a predicted separation of 70 metres leads to almost exactly your 8 sigma event (I got 6.5 sigma, but whatever). So if you think LeoLabs' numbers were credible for an outfit with 30 people, what's hard to believe about the US military being able to track satellites twice as accurately?