Thursday, October 1, 2015

Thursday Morning Storm Update: A Fork in the Road

Good morning once again everyone. Oh did the plot thicken in the last 8 hours and this forecast did not get any easier. This pretty much sums up the model spread currently..

I would even go as far as saying not only is Joaquin approaching this, but it might have its right turn signal on.

To review here where the 3 scenarios I have been discussing.

  • Scenario 1: Storm hits around Outer Banks area as a Hurricane
  • Scenario 2: Storm trends north which would take the Delmarva region and northern Mid-Atlantic into the main impact zone
  • Scenario 3: European model is correct and it is out to sea

After last nights runs we have to assign a 25% probability to scenario 1, 35% to scenario 2 and 40% to scenario 3. This means the European model's stock is going up because now the other models are trending towards it, not the other way around. 

Here is updated model spread, hence the fork in the road..

Who wants to figure this one out? Good luck!

National Hurricane Centers official forecast..

In all seriousness, this is becoming extremely difficult. On last nights 12 am run of the GFS model we had this doomsday scenario with a Jersey Shore landfall...

Just to switch to this below 6 hours later which is its most updated run now..

Everyone should be happy about this. If the first image verifies then we got big problems. The good news is, that is not likely.

Remember on last nights video the GFS ensemble (tests model for potential error). Well that looks like this now below..

Again, huge spread of outcomes.

So where do we go from here? Well we cannot forget about the European model that from the beginning was insisting the storm takes the right end of the fork out to sea. Last night, it stayed consistent..

Meanwhile we still got a whole suite of models showing the complete opposite with a NC landing hurricane !

This is now where I think we have to weight these models by their historical skill set. The European model is the best forecasting tool in the world in my opinion. It looked yesterday like it was about to swing and miss BUT now that I am seeing a trend towards its solution  not away from it, it raises my attention. The model might be understanding the steering currents and pressure interactions of the atmosphere better. 

As a reminder this storm is getting steered by a low pressure trough to the east, high pressure in the north Atlantic,low pressure to its east, and high pressure again to its southeast. They are battling to pull the storm in different directions. You can see those currents below (I labeled them). Think of the storm as a leaf in a stream. What ever model is handling the interaction of all these currents the best is the model that wins...

The only thing to do now is wait. I can't try to be a hero and make a bold forecast because the stakes are too high with a monster Cat 3 hurricane that currently looks like this (storm has strengthened significantly in last 12 hours). It is different in the winter when impacts from snow don't compare to a hurricane. 

I will say this....IF the the European model holds its ground today AND the other models continue their trend towards it, it is game set and match in my opinion-we will not have this massive storm hit the east coast. However if this spread continues then the middle ground of scenario 2 might start to gain more merit. 

By the way here is update for Friday's rainfall amounts..

Notice the sharp northern cut off. That is due to strong high pressure. It looks like rain can clear out by mid morning Saturday prior to arrival of potential main storm Sunday into Monday. 

Thanks for following guys, and I will do my best to keep everyone updated on this evolving difficult forecast situation. I will try to nail this down soon but given the model spread I need to wait a little longer. 

What will today's model runs bring? Will we get more clarity towards one of these 3 scenarios? Find out tonight in my evening update!

1 comment:

  1. Good post, Willy. Has anyone ever done a backtest on the statistical accuracy of the various models?