Because by international treaties signed by the US, the country that launches it’s space hardware must take responsibility for that hardware until it is de-orbited. Because the spacecraft is out of control the US must acknowledge this fact. But I will tell you that they can predict relatively certainly where, when and how it will come down.
Only after the final phase of descent starts. It's like a roulette wheel. For example. you are driving a car on a hilly, uneven terrain. The engine suddenly stalls. Can you predict where the car stops?
The answer is "No" because it depends on your initial speed and on the incline of the road (all the way to the stopping position) and on the wind, and on the type of pavement (concrete vs. sand) and on many other factors.
In case of a satellite it all depends on the early atmospheric drag on the descending shape. Initially this drag is miniscule, but it depends on which way the satellite is oriented (not known) and how high the upper atmosphere reaches right now (not known). But this microscopic difference early in the descent can make the satellite fall in South Africa or in North America. Only when the satellite is on its last spiral around the globe, when it is already burning like a meteorite, then it is reasonably clear what will happen because there aren't that many options left.
Controlled landings are far more exact, but only because the spacecraft is precisely controlled during the descent, and the descent path itself is designed to dive deep into the atmosphere.
Well, HOW certainly, and WHEN to maybe within a few days, which is more or less implied by "late February or March" ( if we assume "early March",) but to know WHERE you have to know WHEN to within a minute or less, since it's moving at about five miles per second. There is enough variability in the density of the upper atmosphere to make it impossible to predict the moment of final descent with this kind of precision. It's like predicting when a picture will fall off the wall.
That’s comforting. The fact they are talking about it at all, and talking about “mitigating damage” as to where it will come down, makes me think that it might come down in a crowded place.
I rather doubt that.
That's news to me. I have always heard that the upper reaches of the atmosphere are more like (changeable) mountain ranges than plains. That being the case, it is impossible to say when - which particular orbit - it will begin to impact atmosphere, how much (or how little) it will hit, etc. until it is beginning to significantly slow and de-orbit.
NASA and others that watch orbital and sub-orbital objects can predict re-entry points for things like the shuttle or ballistic missiles with more accuracy because they are coming down at higher angles than something nominally still in orbit, just brushing the atmosphere. In the case of a dead satellite, I'll bet they can't even say with certainty which orbit it'll start hitting air on, let alone which part, how much, etc. If it were really important to them, they'd save a little maneuvering fuel and at a good time, nudge it down to give it a predictable re-entry - one steep enough to guarantee burn-up, or splash down in an uninhabited place (eg. ocean) where any parts couldn't be recovered.