Erika forecast tracks, color-coded by intensity, from the 06Z GFS ensemble run. The 00Z ECMWF run is similar on Aug. 27, 2015. (

The forecasting and coverage of Tropical Storms Danny and Erika revealed a host of problems in how meteorologists and the media are communicating about tropical weather systems, and how the National Hurricane Center is missing an opportunity to lead the conversation.

Particularly in the case of Erika, there was an inordinate amount of hype and bad information disseminated for a storm that dissipated before reaching the U.S., which could’ve been quashed if the National Hurricane Center had a stronger voice.

Some leading meteorologists are doing some serious soul-searching about how to address the situation. I want to specifically point to and praise essays written by Nate Johnson, a broadcast meteorologist out of Raleigh, and Bryan Norcross, senior hurricane expert at the Weather Channel, which articulately describe some of the problems and propose solutions.

Below is a synthesis of their comments along with my own two cents.

1) We are overselling single computer model forecasts

As Erika was organizing in the tropical Atlantic, we saw lots of scary forecasts of the storm striking Florida originating from single computer model simulations.

“The meteorological hazard of highlighting a single run of a single forecast model should be obvious,” Johnson writes. “Models are guidance, not forecasts. They are subject to a multitude of errors. … As a result, a given model may be subject to wild swings from one run to the next, owing to the interplay of these and other factors. The old saying goes, ‘All models are wrong, but some models are useful’, and there certainly were a lot of ‘wrong models’ with Erika.”

To illustrate his point, Johnson shows a compilation of the various model simulations for Erika with its actual track superimposed (via Michael Lowry, tropical weather expert at the Weather Channel):

Here’s Johnson’s punchline: “Featuring a single model run, especially one that features a high-impact outcome, can drown out communication of other, far more likely scenarios, including the official National Hurricane Center forecast or even the consensus of better-performing models.”

To overcome the problems with individual models, Johnson recommends that weather communicators feature “spaghetti plots” which include the track forecasts from a whole group of model simulations that give a fuller range of a storm’s possible track.

But spaghetti plots on their own aren’t enough, Johnson stresses. He suggests meteorologists add value to this kind of model output by explaining which specific forecast scenarios, conveyed by the spaghetti plots, are the most and least likely. This requires some expertise and knowledge of which models work best in different situations. He points to this tweet from Bryan Wood, a meteorologist from the insurance industry, as providing a great example of model output with value-added expert interpretation:

2) Some media headlines are overselling the storm threat

News organizations, to attract clicks, will jump at the first opportunity to headline the possibility of a landfalling U.S. tropical storm or hurricane, sometimes even when it isn’t justified.

When AOL headlined a story titled “First hurricane of 2015 might strike U.S. shores soon” after Tropical Storm Danny formed, I wrote it was “desperately misleading” and that such over-the-top headlines may cause consumers to lose faith in weather predictions.

[Dial back the hype: No, AOL, a hurricane isn’t striking the U.S. ‘soon’]

There’s no easy answer to this problem. Maybe we call out bad, irresponsible headlines when they appear and hope the offending news organizations dial them back. However, such policing may devolve into a never-ending game of whack-a-mole.

Perhaps, the best strategy is for responsible media to simply stick to its mission of providing compelling and credible information, and let the cream rise to the top.

3) The National Hurricane Center is not reaching the public and media with clear, consistent messages

During Erika’s life cycle, uncertainty plagued every aspect of the forecast. The National Hurricane Center stressed this uncertainty in its technical discussions, but such information was not front and center in its public advisories and other products.

One wonders whether Florida would’ve declared a State of Emergency for Erika, which in hindsight looked silly, if the responsible decision-makers knew there was a very real chance the storm would fall apart.

The Weather Channel’s Norcross writes the Hurricane Center’s antiquated and inflexible product suite is preventing it from presenting the most important messages to the masses.

“Only the most experienced user knows how to aggregate an accurate understanding of the NHC’s thinking by assimilating the multiple links and formats, and even then the bottom line is often buried in the verbiage and/or a technical understanding of the products,” Norcross writes. “Most media people, social and otherwise, are not experienced users or are overwhelmed by the mechanics of deciphering and communicating the message on deadline. The result is: the public suffers.”

I have long felt the Hurricane Center should, for all active storms, have an illustrated public summary that, using plain English and easy to understand graphics, leads with the key storm messages, and then follows with information about forecast confidence, different scenarios and the range of possible storm impacts.

As the Hurricane Center is probably the most trusted source for hurricane information (and rightly so, given its expertise), it should be the organization guiding and even controlling the message during active storm threats.

Frankly, if the National Hurricane Center had a stronger, more accessible voice, it would drown out the clamor from misleading models and horrible news headlines in digital media.