Weather forecasters can’t manipulate hurricane warnings — here’s why

Damage from Hurricane Matthew in North Charleston, South Carolina, October 2016. Ryan Johnson/Flickr, CC BY-SA

Many mini-dramas develop during major disasters like Hurricane Matthew, which has left a trail of devastation in the Caribbean and the southeastern United States. One such drama occurred outside of the storm zone: Conservative news blogger Matt Drudge accused the federal government last week of hyping the threat to the U.S. coast, purportedly to play up possible links between extreme weather and climate change.

The Twitter universe rightfully jumped on this claim, calling it extremely irresponsible. Some critics invited Drudge to stand on the central Florida coast and witness the passage of Matthew so he could personally verify its strength.

But here is a larger, key point: It would be virtually impossible for our government, or any weather service, to intentionally overhype or downplay the real risks of a major hurricane as it approaches the United States.

I have been involved with operational weather forecasting for nearly 40 years. From 2005 through 2009 I was responsible for typhoon forecasts that the Defense Department’s Joint Typhoon Warning Center, or JTWC, issued for the western Pacific and Indian oceans. After retiring from the Navy, I served as chief operating officer for the National Oceanic and Atmospheric Administration. In this position I was responsible for the National Weather Service and its components, including the National Hurricane Center, or NHC.

During my career I have seen the internet and social media transform weather forecasting in general, and hurricane forecasting in particular, from a skill that was seen and practiced by a small community of specialists into one of the most transparent technical endeavors we do on a routine basis. Every forecaster I have ever worked with – military or civilian – wants to get predictions right. And even if they wanted to shade the forecast one way or another to support some agenda, it would be impossible to do it in today’s networked world.

The weather community opens up

In the pre-internet era, hurricane forecasting was more of an art than a science. Modern weather forecasting developed in the 1960s with the advent of radar, computers and satellites. Well into the 1980s, forecasters were still working out how best to integrate satellite data into computer-based forecasts, and both satellite measurements and computing power were crude by today’s standards.

Observations and computer prediction models traveled on government-only, proprietary circuits, with only limited access by private forecasters or academia. Government forecasters would share storm track and intensity forecasts with the public, along with short written discussions (TRANSMITTED IN ALL CAPS), but little else. The actual processes for deriving forecasts were closely held, available only to members of a very select and specialized guild. Private-sector forecasting of hurricanes was in its infancy, hampered by a lack of real-time information.

Doppler radar image of Hurricane Matthew over Florida, Oct. 7, 2016.

Starting in the 1990s personal computers, dial-up access and then the internet fundamentally changed the way weather information could be accessed and distributed. Today global weather models from the United States, Canada, United Kingdom and the European Centre for Medium Range Weather Forecasting are available to anyone with an internet connection. When the NHC directs U.S. Air Force Reserve personnel or NOAA “Hurricane Hunters” to fly a hurricane reconnaissance mission, the data they collect are posted virtually in real time. Satellite images from the JTWC or NHC forecast are available online free of charge.

The International Space Station tracks Hurricane Matthew from low earth orbit on Oct. 4, 2016, taking images and video of the storm. NASA/Flickr, CC BY-NC

For hurricanes, the federal government produces one official forecast from either the NHC for the eastern Pacific and Atlantic oceans or the JTWC for the western Pacific and Indian oceans. This forecast draws on computer-based weather forecast models, an assessment of the storm’s real-time characteristics, and the knowledge of a highly trained typhoon duty officer or hurricane specialist. It would be obvious instantaneously if the forecast deviated substantially from observed conditions or from a blend of the computerized forecast guidance without providing some meteorological or physical explanation.

This level of transparency is fairly new. As recently as the early 2000s, there were significant and sometimes emotional debates within the weather forecasting community over how much computer modeling data and information from weather observations should be publicly available in real time. Some forecasters worried (and still do now) that users could misinterpret individual pieces of data or second-guess official forecasts. Over time, however, consensus has grown in favor of making all the data available to anyone interested, so that everyone can see how forecasts are put together.

Forecasting hurricanes in the internet era

Now that so much weather data is publicly available, the explosion of social media guarantees that thousands of observers are looking over forecasters’ shoulders. And forecasts get a lot of attention, especially when they involve extreme events.

Hundreds of people with varying credentials comment on every tropical storm, hurricane or cluster of thunderstorms. Popular sites like Weather Underground and Windyty further attest to broad public interest in all things meteorological.

Poster from Weather Underground. Weather Underground

If the NHC or JTWC appeared to be disregarding either observations or reliable forecast models without explanation, weather enthusiasts would quickly point this out on social media, and major news media would pick up the story. We don’t see this in headlines because it doesn’t happen.

With a few days of hindsight, it’s clear that the NHC was eye-wateringly accurate in its forecasts of where Matthew would be two days in advance, and pretty good at forecasting the storm’s position, three, four, and even five days in advance. NHC initially forecast that the storm would remain east of the U.S. coast in the Bahamas. Gradually the storm track “walked” westwards as the computer forecast guidance trended in that direction. Matthew also traveled close to the coast farther north than initially forecast, producing extreme rainfall in North Carolina (which can be physically linked to a warming climate). This aspect was predicted with impressive precision up to five days in advance.

National Hurricane Center map of Hurricane Matthew’s path and projected course on Wednesday, Oct. 5 (click for larger image). National Hurricane Center

Weather forecasters understand that they walk on a knife’s edge between overwarning on the one hand – which can produce false alarms and complacency – and underwarning on the other, potentially trapping people in life-threatening situations. The NHC, backed by a national and international network of observers, scientists and computer models, did an amazing job on a very dangerous storm. There always are lessons to be learned, but the bottom line is that we owe them thanks and Mr. Drudge owes them an apology.