It was about as close to zero as you can get. Was it ZERO? No. But by advisory #33, at 11 AM on 1 Sept, the 6Z models had come out and every one of them had the recurve scenario. Even the ensembles of the global models had him recurving. There was not ONE model that put Alabama in ANY danger of damaging winds...not a single one.
I spent 32 years as a professional meteorologist....most of that time I spent it forecasting tropical cyclone impacts. It was a good forecast to say NO impacts give every shred of data available. Any forecaster that would have said "impacts were possible" would have just been playing it safe or playing politics and not making a real forecast....because there would have been ZERO data to support such a conclusion.
The uncertainty levels are still far too high to blow off the chance from a risk management perspective.
Downside to prepare? Very little. Downside if not prepared? Huge. Any risk manager is going to say yes if you’re in Alabama you take basic preps for that Cat 5 headed in your direction, even if the odds are high it will turn away.
The logistics of disaster preparation are an exercise in risk management, whereas forecasting is not.
If you were in Puerto Rico and listened to the forecasters, you got a false alarm. Not one forecast put Dorian east of PR, yet there it went anyway.