This is quite exciting. In my decades as chief ATC Coordinator for United Airlines I consistently argued that the excuse of “unnecessary and unrecoverable” delay used to wait out a TMI decision point was a mistake. Of course this argument was usually espoused by carriers who would take little of the delay if we lost the wager. I’m also interested in whether your model uses ITWS type of data to forecast the impact of low level winds on approach that can cause compression on final. And lastly, it might be interesting to infer quality performance patterns by ATC as related to controllers, front line managers, and TMU specialists work schedules.

Expand full comment

We haven't used ITWS yet, given it's relatively short forecast horizon. It's funny you mention LLWS—that's been on my list since the start, but didn't make the cut for our MVP because it requires working with a relatively unruly, 3-D version of the HRRR. Now that we've launched and we're circling back to that cut list, perhaps we'll explore forecasted wind at various isobaric levels.

There is some seasonality, though not confident it's so good as to pick up on skill of individual controllers/managers/specialists (maybe staffing outage patterns).

Expand full comment