From BMX:
Let's talk about the uncertainty...
We have several different models we examine for weather. Each model is initialized differently, putting different weights on different parameters. So when the models run their algorithms, because they initialize differently, we can get widely different results.
When models are in agreement (generally), or have consistency, for many days out in advance before a big, impactful system, this tends to lead us to higher confidence. If models, generated differently using different weighting and different algorithms, agree, then surely that means the event will occur....right?
Sometimes.
Every now and then, one of the models picks up data that causes it to produce different output. This could be wonky data that gets ingested and causes something different. Or it could be that the model really is picking up on something in the pattern that the others aren't catching yet.
Up until last night, the low pressure causing all this rain was expected to be across the state, which would cause plenty of cold air for the rain to change to freezing rain, snow, and sleet. Last night, one of the models starting showing a vastly different solution, and pushed that low much farther north. This would prevent enough cold air from changing the rain to the snow/sleet/freezing rain. This would change the impacts significantly.
We're not advertising this widely for many reasons:
1. It JUST changed, so our confidence is low. We would need to see if the other models fall into agreement, or that model moves it's way back to what it was showing.
2. If we start advertising "little to no impacts" and then the model changes for the worse, that is MUCH harder to advertise the worsening conditions than to go the other way.
3. Looking at the environment, there could be enough cooling as the precip falls, or other environmental considerations, that colder temps could be seen, colder than what the models are showing at the surface.
So what can you do? Stay aware of the weather and know that this is a complicated system (winter in AL is by default complicated). We'll update you with any and all changes. And know we're watching every data ingestion, every raw data product, sounding, graph, chart, and measurement like a hawk.
If the low moves north and we end up with mostly rain (and maybe some storms), then consider this an exercise in preparation. If the low remains south enough to experience wintry weather, then we'll have some advance notice.
