Keeping an Eye On Matthew

As I've mentioned before, keeping track of what goes on in the tropical weather is a late summer activity around here.  Not quite a hobby, but it gets looked at on regular basis during storm season.  I'm sure most of you have heard about hurricane Matthew, which underwent an impressive intensification today up to a Category 4 storm with 140 mph winds.  The current official National Hurricane Center prediction for the next 5 days, looks like this:

This is an unusual situation, with the fairly hard turn to the north.  The various models have been predicting this turn for a few days, and the system has been behaving like the models.  It looks like the eastern tip of Jamaica is likely to get hit with Matthew at this sort of wind speed.  At this point, it's hard to tell just where it's going to hit, but it's looking like Kingston area will be spared the worst of it, and Port Morant and the relatively lightly-inhabited east end of the island will get the worst.

That's IF the models don't break down.  There are some unsettling things going on with this storm.  The models seem to have diverged into two sets.  Each set is generally considered a good quality set, but the two of them don't really match.

I clipped this graphic from the Central Florida Hurricane Center, an independent place (that is, not affiliated with any of the major weather companies or services) where a group of weather geeks hang out to get updates and discuss the storms.  This link goes to the Hurricane Matthew lounge - a discussion forum.  Getting back to the two plots, notice how much wider the "hairball" of predicted tracks is on the left plot than the right?  The left group is the ensemble of models from the ECMWF,  the European Center for Medium-Range Weather Forecasts;  the right group is from the GFS, the Global Forecast System, which is US-based, not European.  Just pay attention to the area bounded in yellow; the other tracks and loops are from systems that may or may not even be out there, yet.  The GFS is tightly packed, keeping the storm off Florida and pushing it north, where it could conceivably hit the Atlantic states from North Carolina all the way through Maine.

Why so different?  Is one set of models better?  In general, no.  They're both well-trusted models and neither one is "slam dunk better" than the other.  This raises the possibility that there's something wrong with one of them.  When other evidence is considered that includes the big picture weather fronts and systems, the more scattered ECMWF forecast may be a better model run.  The tight clustering of the GFS ensembles a few days out could actually be an indication of a systemic error in that model.  Something about the way the weather develops causes the models to break down.  Since a healthy percentage of those ECMWF tracks put a major hurricane on my head within a week, I can't say I like that idea!

Hurricanes, monsters though they might be, are driven by the weather systems around them.  It's like two dissimilar fluids in the same container.  Think of a lava lamp.  The moving lighter colored fluid moves the darker fluid around.  Thinking of it that way, what would make the storm do a right turn like that?  When a large system rotating counterclockwise gets in its way and the winds push the storm north.  The other word for "large system rotating counterclockwise" is a low pressure system, and there currently is such a system over the southeast US.  Just offshore the east coast is a high pressure system, which rotates the other way.  Somewhere around midway between both systems, the counterclockwise winds from the Low and the clockwise winds from the High combine to form a highway to carry Matthew somewhere north of where it is now.

The difference between these models could be how accurately they predict what that high pressure and low pressure system do.  One of the forum members posted this prediction of the way the steering currents look tomorrow morning. 

If you look between Florida and around Bermuda, you can see a low pressure system centered over Kentucky and what I'll call the Bermuda High (it's a semi-permanent high in the vicinity of Bermuda).  They show a clear, narrow path up over the Carolinas, then the Chesapeake/Delmarva area possibly into New York.  Of course everything on that map is going to move long before that, but notice that Matthew's path into that "conveyor belt" north is being blocked by winds pushing it to the south and southwest.  It may not be picked up at all, or not for several days.  That would be more in keeping with the ECMWF models than the GFS. 

That implies the National Hurricane Center is wildly off in their forecast.  It happens, but it's safe not to bet your lunch money on it.  At this point, all we can do is watch and see what happens.  This is where you smash your fondest theories against reality and see if you're able to predict what's going to happen.  There's hardly anything more fun than that.