Hah! And just like when I was still an active researcher, I did a first-pass literature-search to check if anyone has tried doing this very thing I've been thinking of.

AKA: "Real-Time Detection + Forecasting of Turbulence Regions from ADS-B Data"

The good / bad news:
* Hah! As is usual for research-y projects I think of working on, this is actually something that has been tried... and/but

* Yay! In this instance actually, they actually managed to get something working along these lines [1] - though with a few additional parameters I hadn't thought of using, but very much trying to do what I was thinking of...

* The bit that makes me feel less like an outdated/hopeless dinosaur though are:
a) It sounds like they are apparently only still in the "trial" / "validation" stage currently, from the press releases I saw.
b) It's "still" in that stage, as they only really announced that they got it working late last year / early this year! (For context: Usually, I end up rehashing things from several decades ago instead! So to only be almost a year off the pace in this instance is really not too bad, all things considered 😜)

[1] https://aireon.com/using-ads-b-data-to-detect-turbulence/