Well, it kind of depends, right?
I don’t imagine it’s going to be an analog constantly calculated CFD weather state… I’d imagine they’d calculate a state say every 2 hours or something, using the MeteoBlue prediction as the basis an hour or two out, then calculate how fronts move between those two data points to simulate the motion of a front over time?
It’s an interesting question, how do you simulate a moving front?
I wonder what the discretization size of a box they’d be caluclating within will be, and how do they average at the boundaries of the boxes?
The whole point is going to be calculating how air is interacting with the terrain and flow across an area, and they’re going to be calculating thermals over an area… boggles the mind how much data they are going to create.