The models are locked in cement right now. Right now, most models send Irma up Florida's east coast, but NVG is a bit of an outlier, sending Irma up Florida's west coast. Saturday will be the BIG DAY, when Irma starts turning north for sure, or delays turning. I'm thinking it will turn early (but maybe very close to Miami). If a turn is delayed, it could be very bad for Tampa.
I tried to make some crude annotations to model output. The figures show the 500 millibar (mb) pressure level heights above sea level. The units are meters. 5,800 meters is about 19,000 feet above level. Half the mass of the atmosphere is above 19,000 feet, and half below. A ridge is where pressure is higher than average and where heights of the 500 mb level are higher than average. A ridge looks just like that on these weather maps. Similarly, a trough is where pressure is lower than average and where heights of the 500 mb level are lower than average. A trough looks just like that on these weather maps. The colors refer to vorticity - if you will, the 'swirliness' of the atmosphere. The atmosphere is rotating counterclockwise where the colors are orange, and rotating clockwise where the colors are blue.
Illustrated below is the NVG model output - the Tampa Catastrophe Scenario.
Figure 1 is the weather situation this morning (Thursday, 6 a.m., Sept. 7th).
I've illustrated wind flow with crude arrows. Wind flow is along the lines of equal height of the 500 mb surface. (These lines are called isohyets). Where these lines are close together, where the height gradient is greatest, is where wind speeds are the highest. The looping band of high wind speeds across the top is the Jet Stream. The Jet Stream acts like a meandering river, and will form the equivalent of river oxbows when it encounters unusually slow-moving air. One of these situations is illustrated off the Oregon coast. The trough is 'digging' here, and will form what is called a cutoff low - a low pressure system spinning in isolation from the Jet Stream. The low has been 'cut off' from the westerlies.
Usually cutoff lows form from the Jet Stream. The Jet Stream spots more sluggish air to the south and injects the cutoff low into a vulnerable ridge. In the subtropics, though winds flow from the east. The tropical easterlies are more sluggish than the Jet Stream, and rarely do they have the power to inject a cutoff low into more sluggish air to the north. NVG is suggesting that unusual scenario might occur.
The weather situation in the U.S. is quite unusual right now, with a very strong ridge in the Plains and Rockies and a very strong trough along the East Coast. The amplitude of the wave is very large, and can't last long.
There is a small, weak low pressure system present in the Dakotas that is eroding the Plains ridge. NVG says it makes sense for the Dakota low and Hurricane Irma to start rotating around a common center in the lower Mississippi Valley, forming a cutoff low. That common movement, similar to dancers linking arms and swinging around each other at a square dance, is what makes Irma move north so abruptly.
Figure 2 is NVG's modeled situation (Sunday, noon, Sept. 10th). The Oregon trough is now completely cut off and spinning over southern California. The big trough is also cutting off from the Jet Stream near Maine. The ridge over the southern Rockies is now very weak. The weak Dakota Low spins off a tiny band of vorticity (marked with orange colors) which arrives at Irma. The subtropical easterlies sense the weakness over the southern United States and decides to use Irma to inject a cutoff low into the weak ridge. Irma suddenly shoots north to Tampa.
Figure 3 is the NVG modeled scenario (Wednesday, 6 a.m., Sept. 13th). The California and Maine cutoff lows are no longer isolated from the flow, and have rejoined the flow of the Jet Stream. Irma has crushed Tampa and Irma's remnant has moved into the lower Mississippi Valley, forming a cutoff low in the middle of the ridge.
That is the nightmare Tampa scenario.
Personally, I don't like the NVG solution. Hurricane Irma is strong and the Dakota low is weak. There is no compelling reason for them to work in concert. The Dakota low isn't powerful enough to fling Hurricane Irma north. The GFS solution is still the most likely one - the storm stays east of Florida and the storm slowly moves north. And it's still possible there is no northward movement at all and the storm just keeps heading west.
Meanwhile, the less-imaginative GFS model suggests Irma will get captured by the East Coast trough in a standard manner, and that's why it moves the storm north.
And there's a third possible solution: Hurricane Irma keeps moving west without a turn, following Hurricane Katia, and ends up on the Mexican or Texan coast. That's a different way to make a cutoff low, and quite a logical solution too.