How might an "imminent Starlink reentry in your area" alerting script work?

Question

The video below Space debris over Puerto Rico (2/7/2022) SAC seems likely to show one or more reentering Starlink satellites from the 40 "Starlost" lost to atmospheric heating due to a geomagnetic storm.

• Why didn't SpaceX see that geomagnetic storm coming? Was this a fluke or could this happen more frequently in the future? (R.I.P. 40 lost starlinks)

Wouldn't it be great if one could generate alerts to reentering Starlinks so more videos of those pesky astronomer-foiling Starlinks experiences its fiery demise can be recorded?

Question: How might a Starlink reentry alerting script work? Besides scouring TLEs and looking for low altitudes, what might a script look for either within the TLE or from other sources to get an idea that a Starlink is in imminent danger of burning up?

How might an "imminent Starlink reentry in your area" alerting script work? Besides scouring TLEs for very low altitudes and SGP4 propagating below the Karman line, what could it do and what other sources of information are available?

See also

• The extensive notes below the YouTube video
• Se observa basura espacial sobre el cielo de Puerto Rico Expertos aseguran se trata de satélites Stalink lanzados el 3 de Febrero de 2022

Answer 1

Well, I can't give a great definitive answer, but I can at least relate my experiences writing my own Starlink reentry analysis script.

There are several challenges that make it difficult to create a script that produces even remotely accurate results:

1. TLEs are bad for propagation as the spacecraft dives deeper into the atmosphere. The rapid change in drag forces means that a TLE fit to "yesterday's data" will not extrapolate well into the next day. You have to take a TLE state and convert it to a full numerical integration.
2. Lack of knowledge of Starlink de-orbit maneuvers. Obviously a Starlink's reentry path is heavily influenced by any de-orbit maneuvers to speed the descent, and SpaceX doesn't communicate them. (although that's not an issue for these particular Starlink reentries)
3. Lack of knowledge of the Starlink attitude as it descends. Its attitude profile will obviously change the drag force it experiences, and it's also unclear at what altitude the spacecraft loses attitude control and begins tumbling. That is a function of the amount of torque the ACS can deliver, which is also unknown (at least publicly).
4. Difficulty predicting space weather. Variability in space weather makes it difficult to predict drag forces, and therefore reentry times/locations. Often this is boxed with high/low drag predictions, or else some sort of Monte Carlo is applied to give a range of possibilities.

So, with that all being said, I did the following:

1. I downloaded TLEs from Space-Track, used a high-fidelity force model to fit a numerically integrated state to that TLE
2. I used that fitted state to propagate forward, under high/nominal/low drag predictions and marked the lat/lon locations of the spacecraft as it passed down from about 95 km to 80 km altitude.
3. I also did a check if the locations were in local day or night, as an attempt to predict if the reentry would be easily visible.

For my attempts, I had issues with the fact that those Starlinks were being actively de-orbited, and therefore maneuvering, so my predictions didn't fit all that well.