AP EXPLAINS: How one computer forecast model botched Ian

The National Atmospheric and Oceanic Administration determined on Thursday that one of the major American computer models used by forecasters missed a “critical” error – postmortem computer models.

According to Ryan Maue, a former chief scientist at NOAA who is currently working as a private meteorologist and was not involved in NOAA’s analysis, “The significance of the mistake is quite evident.”

Although it was at times on the farthest edge, the eventual landfall of Hurricane Ian was always within the “cone of uncertainty” of the National Hurricane Center’s forecast, specifically in southwestern Florida. Despite meteorologists’ official forecast, Hurricane Ian did not completely miss.

Ian powered through the Caribbean while computer prediction models, which had previously reached a consensus on the path of Hurricane Fiona, were separated by hundreds of miles. However, the situation was more complex than that.

The favorite simulation model among many meteorologists, which accurately predicted the landfall of a British hurricane in the Panhandle of Florida earlier this year, performed even better than any other model in 2021. This particular European computer model, known for its reliability, outperformed the American model that is typically considered more dependable.

Trying to avoid what meteorologists call the dreaded “windshield wiper effect,” the NOAA official stayed in Tampa, where lots of people are vulnerable to gigantic storm surges that could make the situation even worse, with the possibility of landfalls seeming to be centered on the south.

While many people were worried about Tampa, Ian wasn’t concerned.

Tampa was fortunate to avoid the most severe storm surge and rainfall as it was located to the north of the destructive right-side of the hurricane eye. Although the distinction may seem insignificant, it falls within the 100-mile (161-kilometer) margin of error that NOAA allows for a substantial storm. The hurricane ultimately struck land 89 miles (143 kilometers) south of Tampa, in Cayo Costa.

People wondered why the worst-case scenario didn’t occur. There are meteorological, computer, and communications factors.

Alicia Bentley, from the postmortem agency, stated that the American model was the most accurate in terms of strength and timing when it came to tracking Hurricane Ian. However, the British model performed slower and had the closest eventual landfall in Florida. On the other hand, the European computer model performed the best overall.

Brian Tang, a professor of meteorology at the University of Albany, stated that during Ian’s five-day track, the average error in the American model’s calculation was closer to 520 miles (325 kilometers), whereas the European model’s error was closer to 350 miles (220 kilometers).

During an interview, Tang mentioned, “Much of what catches our attention in the general public is when significant errors occur, and these errors have an impact on individuals residing in densely populated regions.”

The Fort Myers region experienced the full force of the storm because individuals who left Tampa may perceive it as not technically a near miss.

In the last 10 years, the official three-day forecast error has been reduced by half, from 148 miles (92 kilometers) to 278 miles (172 kilometers). Some people are spoiled because the average error in hurricane track forecasts has improved significantly.

According to Tang, the European model has shown improvement, just like the American model, following a significant increase in NOAA funding. Although the American model is released later and takes more time to process, it is also more intricate and utilizes a greater number of observations. This is because meteorologists have been promoting it as superior for many years.

Brian McNoldy, a researcher at the University of Miami, explained that the models employ a comparable physics equation to replicate atmospheric processes. Typically, these models depend on similar observations, with some variations. However, their divergence lies in the way these observations are incorporated into the computer models, the inclusion of uncertainties, and the commencement time of the simulation.

“You are assured to conclude in a distinct manner,” McNoldy stated.

He said, “that makes you anxious,” while Ian was pointing out how different they are if the problem is not similar to the models shown.

According to Kerry Emanuel, a professor of meteorology at MIT, individuals often mistakenly prioritize the funnel-shaped cone indicating the projected path of a hurricane, rather than considering the potential impact in specific areas. Both Emanuel and McNoldy advocate for the removal of this line, suggesting that individuals should only focus on the central line within the cone, rather than overlooking the overall perspective.

Another problem meteorologists face is that outside the cone, storms like Ian, which are big and have high winds and heavy rain, can easily impact areas with a radius error of 100 miles (161 kilometers).

Gina Eosco, who leads a NOAA social science program aimed at enhancing storm communications, stated, “The cone was never meant to depict the real effects. Its sole purpose was to depict the paths.”

Eosco stated that this will aid the organization in determining whether it needs to modify its cautionary communication. Therefore, NOAA conducted a survey of individuals residing in Florida, Georgia, and South Carolina prior to Ian’s arrival, and will subsequently conduct a follow-up survey to assess the public’s perception of the hazards conveyed through media and government sources.


Stay updated on AP’s coverage of climate and environment topics by visiting https://apnews.Com/hub/climate-and-environment.


Follow Seth Borenstein on Twitter with the handle @borenbears.


The AP is solely responsible for all the content. You can find more information about AP’s climate initiative here. Numerous private foundations provide support for the Associated Press’ coverage of climate and the environment.