MWN Blog: Commentary: On forecasting tropical systems that haven't yet formed

You've likely heard the hype. A tropical wave that has not yet formed into a storm is "bearing down on the Caribbean" and sure to pose the gravest threat to the United States since Twitter was born. Millions are in the potential path! Maybe you've even heard of a "triple threat" to the U.S. because there are two other storms in the Atlantic, even though they have virtually no chance of reaching American shores.

Tonight I thought I would take a few minutes to explain why responsible meteorologists (as I consider myself to be) can't, and won't, predict what the yet-unnamed tropical cyclone approaching the Caribbean is going to do. It's part of the education process I engage in with MemphisWeather.net, particularly because I display the Weather-Ready Nation and NWA Digital Seal logos that mean I am a responsible purveyor of weather information.

But first, why even bring this up? We aren't near the coast, so why does it matter? Because I get your questions and answer almost all of them! "What does the European model say today? It's always right!" Also because many of us have interests in near-coastal locations - relatives, friends who have moved, or maybe even our favorite beach spots. The Mid-South is also near enough to the central Gulf coast that when a big storm heads that way, we need to pay attention. If it doesn't directly affect our weather, it likely will indirectly.

Forecasting in the early stages of development

So here's the problem. A tropical wave that is still a few clusters of thunderstorms, with no "closed circulation" (meaning it is not yet an actual low pressure system), is not well-observed by the computer models. And if it's not well observed, it isn't going to be well-forecast. There are simply too many variables, including whether or not it will actually form and then how strong will it be and where will it go. Models, and the human "interpreters" that forecast these systems, are getting better, but it's still very difficult especially in the early stages of a system.

Some examples

Look at these examples from today. Here's this evening's model "spaghetti plot." It looks like a pot of noodles! The colored lines are forecast tracks of this particular system by various computer models. The tighter they are together, the higher the confidence in the track as they all have roughly the same solution. The farther apart they diverge, the lower the confidence, because the solutions are very spread out. In this case, a couple models say the system will miss the east coast altogether while a couple point it towards the mouth of the Mississippi River. In other words, the "spread" of the models doesn't tell us who should be preparing for the storm! Sure it looks like southern Florida has the best chance, but when? And how strong will it be? And then where does it go??

The next graphic shows the forecast strength of the system, going out in time. What does it tell me? That we could have a system that remains a weak tropical depression with little impact other than rain for the next week, or we could have a category 4 hurricane in 5 days! The models don't have any idea, even if you throw out that aggressive GFNI model.

What does history tell us about how well the forecasts from the National Hurricane Center (you know, the "experts" at this stuff) pan out? See below. Forecast error is on the left (vertical) axis and time is along the bottom. Each colored bar represents a certain type of storm based on intensity. As you can see, the weaker the system being forecast, the worse the track error is, especially as you get further out in time. The average error for a weak system at 5 days is over 350 miles! And these are the experts that interpret all the data, not the models!

The point here is that we don't know where "99L" or "Hermine" will go or how strong it will be yet. Or even if it will become Hermine for that matter. What we do know is that it's something we, as meteorologists, need to keep a keen eye on. In addition, your responsibility is to keep in touch with your trusted sources and not fall for the bait of a bold headline, slick graphic, or doomsday prediction. Here's how I handle long-range forecasts of impactful weather events, including tropical tracks, beyond more than a few days:

The only time we post longer-range forecasts is if there is A) some model consensus & B) it's within about a week. Even then, with caveats.
— MemphisWeather.net (@memphisweather1) August 20, 2016

Thanks for making MemphisWeather.net one of your trusted sources! I'll continue to work diligently to ensure that trust is not broken, including saying "I'm not sure" if I'm not.

Stay tuned to our social media feeds listed below for the latest "truth without the hype" outlooks and forecasts.

Erik Proseus
MWN Meteorologist

----
Follow MWN on Facebook, Twitter, and Google+
Visit MemphisWeather.net on the web or m.memphisweather.net on your mobile phone.
Download our iPhone or Android apps, featuring StormWatch+ severe weather alerts!