How the National Weather Service Is Working to Reduce Tornado Warning False Alarms

A National Weather Service initiative is tackling the issue of tornado warning false alarms and with some promising preliminary results.

(MORE: Tornado Central)

Launched in fall 2015, the Tornado Warning Improvement Process, or TWIP, encourages NWS offices to share insights from local research on how effective and accurate tornado warnings have been in the past and how they could be improved in the future.

We've written about the tricky, challenging false alarm issue before.

On average, roughly 70 percent of tornado warnings issued in the U.S. are false alarms. This means only three in 10 tornado warnings contain a verified tornado within the warned area during the time of the warning. 

But that doesn't mean you should ignore tornado warnings.

On average, about 60 percent of all confirmed tornadoes occur in an active tornado-warned area. In other words, six out of 10 confirmed tornadoes were covered by a tornado warning at the time, a parameter meteorologists call the probability of detection (POD). 

(MORE: Your Odds of Being Hit by a Tornado)

The goal of TWIP is to not only decrease the false alarm rate, but also to at least maintain or even increase the probability of detection, the number of tornadoes within an active warning.

One such TWIP study by the NWS office near Milwaukee examined 91 tornadoes and 142 tornado warnings in south-central and southeast Wisconsin from 2008 through 2016.

The study by meteorologist Ben Herzog, science and operations officer John Gagan and Cory Rothstein from the University of Wisconsin-Milwaukee, presented at an NWS meeting with local media meteorologists May 21, suggested tornado warnings should be reconsidered by meteorologists in southern Wisconsin when parameters dealing with instability, low-level wind shear and the height of the cloud base are below given thresholds.

(MORE: Strange Tornado Paths)

A lack of low-level wind shear  – a change in wind speed and/or direction with height  – fails to create enough horizontal spin of the air needed to then be stretched into a tornado.

A lack of instability in the low-levels of the atmosphere doesn't allow for air to accelerate upward to help tilt that horizontal spin into the vertical, then stretch and accelerate it into a tornado.

A cloud base that is too high is indicative of dry air that would undercut an attempt at forming a tornado.

Applying the thresholds of those four parameters alone would have reduced false alarms by 45 percent in southern Wisconsin, the study found.

However, Herzog cautioned that the thresholds of the four parameters shouldn't be used blindly, as that would have lead to an additional 13 unwarned tornadoes in the same nine-year period, all either EF0 or EF1.

(MORE: The Most Tornado-Prone Counties in the U.S.)

A combination of these environment parameters, radar signatures and any storm reports are typically used by forecasters to issue warnings.

Another study by the NWS in Birmingham, Alabama, found that its false alarm rate plummeted from 83 percent from October 2007-March 2011 to 58 percent from May 2011-May 2014.

The Delicate, Difficult Balance

There's a balance between trying to detect and warn for every single tornado while keeping the number of false alarms low. The more tornado warnings you issue, the higher chance of a false alarm, which increases public complacency.

It is easier for a forecaster to verify a tornado warning for a supercell on days with a potential for long-track, violent tornadoes. Those signatures are usually distinct, straight out of a meteorology case study.

Tornado warning performance and lead times are higher on these relatively few, volatile days, according to a 2008 study by Kelly Keene, Paul Schlatter, Jack Hales and Harold Brooks of NOAA's National Severe Storms Laboratory and Storm Prediction Center.

(MORE: States With the Most Violent Tornadoes Since 1950)

However, as Ian Livingston of ustornadoes.com found, about 90 percent of tornadoes in the U.S. from 1950-2016 were weak, rated EF0 or EF1 and can occur in more marginal environments for severe thunderstorms and/or display less distinct radar signatures.

Often, the environment spawning these weaker tornadoes is only marginally supportive for tornadoes, and radar signatures are weaker or more subtle.

Other challenges forecasters face in these situations include:

• Radar is often unable to see rotation in the lowest elevations near the ground, particularly for distant storms, since the beam gains elevation with distance.
• Nighttime tornadoes are hard to verify as they're happening unless the tornado is illuminated by lightning, power flashes are seen, or damage has already happened.
• Short-lived tornadoes embedded in squall lines may quickly form, then dissipate, and their signatures may be less pronounced.
• Tornadoes in squall lines or high-precipitation supercells may be obscured by rain, and, therefore, hard to identify by spotters.
• Spotter networks may be more dense in, say, Oklahoma, but far less in sparsely-populated areas.
• Outside of the core severe weather months during spring and early summer, the environments are more marginal, the public is less aware, and perhaps fewer spotters available.

As meteorologist Jonathan Belles wrote in May 2017, in an attempt to address the false alarm problem, lead times for tornado warnings have decreased from 13 to 14 minutes at the beginning of the decade, to around 8 to 9 minutes in the past year.

NWS forecasters now wait until a tornado has begun before issuing a warning more often, according to Dr. Harold Brooks, senior scientist at the National Severe Storms Laboratory, in a presentation given at the 28th Conference on Severe Local Storms in 2016.

In essence, this comes down to a question of whether the public can stomach missing a few more weaker tornadoes (EF0/EF1) in order to lessen the overall tornado warning false-alarm rate.

"For some, a tornado is a tornado – regardless of strength  – and the desire is for a warning on anything that has the potential to produce a tornado," said NWS-Milwaukee science and operations officer John Gagan in an email to weather.com. "For others, the desire is to limit the amount of false alarm in an effort to ensure optimal response for warnings when it matters most (e.g. strong/violent tornadoes)."

Gagan, who took part in the May 2011 Joplin tornado damage survey while at the NWS office in Springfield, Missouri, emphasized that zero false alarms and detecting every single tornado are likely unattainable. 

"The goal is a scientific approach to the tornado warning process, focusing on both environmental intelligence and probability of impact."

Dual-polarization radar detecting lofted tornado debris and enhanced radar scanning techniques have also helped meteorologists issue tornado warnings with more confidence in recent years.

An overhaul of the nation's weather warning process, including tornado warnings, currently in development aims to provide more precise warnings with increased lead time to help decision makers and the general public respond. 

What Should You Do?

Again, the majority of confirmed tornadoes in the U.S. had a tornado warning issued for that area affected at the time.

Despite all this discussion about false alarms, you should take every single tornado warning you receive via your favorite weather app, such as The Weather Channel, a Wireless Emergency Alert, over the radio or from local television seriously.

Act quickly and take shelter. You may have only a few precious minutes.

Rolling the dice that the next warning issued for your area is a false alarm could cost you your life.

Jonathan Erdman is a senior meteorologist at weather.com and has been an incurable weather geek since a tornado narrowly missed his childhood home in Wisconsin at age 7. Follow him on Facebook and Twitter.