Skip to main content
ABC News
Foggy Forecasts: Meteorologists Chat About How Tough It Is To Predict A Hurricane

As a follow-up to my post Tuesday on how meteorologists handle significant uncertainty when a storm’s a-brewin’, I decided to get a few of my favorite meteorologists together for a chat focused on the current potential storm: 99L, which will be called Hermine if it ever coalesces enough to earn a name. Although this discussion took place Thursday, all the model scenarios discussed are still relevant as of Friday morning.

Disclaimer: Most everyone in this chat has a Ph.D. of some form in the atmospheric sciences (except me!), but they all speak on behalf of themselves. For official warnings and advisories, please consult the National Hurricane Center or your local National Weather Service office.

This discussion has been lightly edited for clarity and length.

Eric Holthaus: Hey all, I know many of you are very busy with 99L and/or are government and/or have responsibility to private sector, but I’d like to chat about we’re thinking on this storm, what the risks are, what’s most likely to happen, and what people should be doing right now to prepare.

My most pressing question: How do you avoid the temptation to look past the Florida landfall to what might happen in the Gulf? The latest models show the potential for catastrophe, especially considering Louisiana definitely does not need any more rain right now.

Michael Ventrice (meteorologist at the Weather Company Energy): I don’t think we are going to have a clear indication of that risk until we see the wind shear relax [note: too much wind shear, which is the change of wind speed or direction, can tilt developing tropical thunderstorms and tear them apart] on Friday or Saturday. … The models are going to have a hard time predicting the fate of 99L until there is an actual tropical depression.

Ed Vallee (energy meteorologist at AccuWeather): In my opinion, the uncertainty is just too high. The storm hasn’t even formed yet. Would like to see some “ground truth” before moving past initial Florida impacts.

Eric Blake (meteorologist at the National Hurricane Center in Miami): It isn’t exactly like the storm is tearing it up yet or the model performance is stellar — that’s enough to make me wary of long term, not to mention that personal interests in South Florida keep my immediate focus here.

Holthaus: Yep, models usually do a lot better with tropical systems that have a well-defined center. Right now, 99L definitely doesn’t have one.

Ventrice: For track purposes, if we see a tropical cyclone form, 80 percent of our calibrated European Center for Medium-Range Weather Forecasts ensembles track it across Eric [Blake] in South Florida.

[Note: Many meteorologists consider the European Center for Medium-Range Weather Forecasts model to be the most accurate weather model in the world.]


Vallee: Right — how this thing interacts with Florida and the ridge to the north is still being worked out. The European model has been pretty solid in showing South Florida impact, but hasn’t been stellar beyond that for the reasons above.

Ventrice: But we saw something similar to this last year … with Erika. Eric [Blake] and I were talking about this on a private thread … Euro was too bullish with Erika last year, where the American Global Forecast System model was much weaker, and the GFS was a better predictor in that particular instance.

Vallee: Erika was a unique system with respect to Hispaniola interaction, too. Confidence will (hopefully) rise once it escapes the high shear environment and gets away from that island. Basic upper-air analysis points to a westward movement, but outlining potential track/intensity is difficult.

Matt Lanza (Houston-based energy-sector meteorologist and FiveThirtyEight contributor): Being in Texas and in energy, I can’t avoid the temptation … what happens beyond Florida matters more to me. So I’ve been trying to use sound meteorology and what I know works best to try and craft a message. Sometimes you have to make a Day Six-to-Seven forecast for a tropical system, but obviously what I’m telling people is laced with scenarios and caveats.

Vallee: What’s glaring to me is (despite relatively small strength/track differences) the Euro and the Euro ensemble [forecasts] have been steady for the last three days.

Lanza: A lot of times these things get framed as GFS [Global Forecast System, a U.S. predictive model] vs. European model battles … and they are, but it’s so much more complex than that. The models are constantly being tweaked and changed, as Mike points out, to (ideally) improve predictability. I think especially when you’re dealing with tropical entities in their initial stages, it becomes less Model A vs. Model B and more, “What does the meteorology and history suggest happens here?” This is where being a meteorologist and understanding the atmosphere (and, frankly, history too) is important, in my opinion.

Gina Eosco (senior social scientist at Eastern Research Group, focusing on weather and risk communication): From a “What should people do?” point of view, there are actions people can take well in advance of a storm. Buying batteries and canned goods, refilling prescriptions, checking flashlights, etc., are all actions people can take whether or not this storm strengthens in the Gulf. If nothing happens, then they are more prepared for the next one. The temptation should be to remind folks to check their hurricane preparedness kits, not to overhype the what-if scenarios. However, responsible discussion of the future potential of the storm increases transparency, which can lead to more public trust. This discussion is a perfect example of a scientific exchange that calmly points out the possibilities.

Holthaus: Why thank you, Gina!

Vallee: Yeah, analogs [note: using algorithms to find the closest match between the current weather pattern and past storms] and meteorology tend to get lost (especially nowadays, I feel) in the hysteria of social media and technology. But bottom line, especially given lack of formation: History/analogs should be at the forefront, in my opinion, similar to winter storms. We always hear, “Oh, this looks like the blizzard of ’96 based on analogs.” Similar approach should be done now, I think.

Holthaus: So, on that note, here’s what the analogs were showing as of midday Thursday:


Hurricane Betsy (1965) is the No. 1 analog, by this analysis. That’s probably pretty terrifying to folks in Louisiana right now.

Eosco: Fear-inducing, yes, maybe. I think it’s fair to share these types of analogs — but qualify them. Don’t purposely induce fear by showing it, but rather explain that “past history is one of the attributes used to make a forecast. This analog is one of the reasons why meteorologists are so concerned about the potential track of this storm. But past history doesn’t mean this will happen. It simply gives us cause for concern. It is also possible that it won’t hit land at all.”

Lanza: Yes … the idea that it could track toward Louisiana or the Central Gulf is a real issue. And given both their hurricane history and recent disastrous flooding, you have to be very careful how you discuss it. I think what Gina pointed out is important. Really, I think all you can do is be transparent, as open as realistically possible (without overdoing it), and encourage people to double-check their plans and disaster kits.

Vallee: While analogs are a great tool, think it’s important to reference track/intensity of said analogs. Tough to be transparent without hyping, but that’s what makes a good, communicative meteorologist.

Holthaus: Of course, in the spirit of transparency, no model so far is explicitly calling for a Category 4 landfall in Louisiana like Betsy. In fact, most models keep this a tropical storm for the next five days or so.

John Morales (chief meteorologist at WTVJ NBC-6 in Miami): Let me briefly jump into this discussion to say that I don’t see 99L being similar to Erika. Erika’s core was torn up between western Hispaniola and eastern Cuba’s topography. 99L’s core circulation (for what it’s worth) remains over water and will enter bathtub water temps soon. If the shear relaxes, there’s a wide spectrum of possibilities. As Eric [Holthaus] just shared in the intensity plots, that includes anything from depression to Category 5 in the Gulf.

Holthaus: You guys, I started this discussion hoping to be reassured. But now we’re talking about a major hurricane possibly hitting Louisiana. Obviously the Louisiana flooding earlier this month comes to mind, and 99L will probably be a big rainmaker wherever it hits. The Hurricane Weather Research and Forecast model is showing spots in South Florida could get more than 12 inches of rain.

Vallee: I think a lot stems from how 99L develops for rainfall totals, but yes, if this were to aim toward the Gulf Coast, it’s certainly not good for Louisiana. Big “if,” of course.

Also of note re: HWRF, Brad Panovich tweeted a good graphic regarding its error. I find that especially true before the Hurricane Hunter aircraft identify a true low-level circulation, and a tropical depression or tropical storm officially forms.

Holthaus: Yes, wow, so HWRF is really only accurate about three days out. I’ll keep that in mind.

Morales: The South Florida Water Management District has begun lowering water levels from mid Broward through Miami-Dade counties in preparation for storm rains.

Holthaus: That’s good. Obviously, emergency managers in South Florida are in (or very near) go-time mode right now. The storm is supposed to hit on Sunday. They can’t, and don’t, wait on an official designation from the National Hurricane Center before preparing for all possibilities, including a hurricane landfall.

Morales: Last thought: Can’t wait for the change in NWS/NHC rules that will allow watches and warnings before a system is baptized. Would’ve been helpful this time around.

Holthaus: Agreed.

Lanza: Only thing I’d say is just to remind people to try not to get too much heartburn over every zig and zag they see shared by people/media regarding the models. Some situations, you simply don’t know where things are going to end up, and you just have to be prepared to act over a broad area with a plan. Hurricanes are a part of life here, and this won’t be the last storm we get. Be aware, and have a plan.

Eosco: I echo this. Stay hurricane strong by preparing before a storm. One last comment is that social media is a wonderful place to have a dialogue, but we must remember that it is part of a public sphere. The scientific community should discuss all of this, but give care to how it is discussed. Don’t overhype. When the next big one does hit, we want folks to listen. We must maintain a tone of openness and transparency without hyping and inducing fear. Stay calm, be prepared, and be thankful we have so many great experts gauging our hurricane risk. =)

Vallee: Yep, agree with everyone else here. Thanks for the opportunity, Eric.

Eric Holthaus is a meteorologist who writes about weather and climate. His articles have appeared in Slate, Vice, Quartz, the Wall Street Journal, and Rolling Stone.