New rating system charts a path to improved tornado forecasts

All tornadoes — whether small or large — originate from thunderstorms, but not all thunderstorms are the same. Different environments and situations create forecasting challenges. For instance, nighttime twisters, summer tornadoes and smaller events can be tougher to forecast.

Researchers wanted to quantify how much tougher, and have published a new method of classifying tornado environments according to their forecast difficulty.

In a new paper published online in the Bulletin of the American Meteorological Society, University of Washington scientist Alexandria Anderson-Frey, and Harold Brooks from the NOAA National Severe Storms Laboratory describe a new way to rate and possibly improve tornado warnings.

“With this research, we’re trying to find ways to truly level the field related to the difficulty of the forecast situation,” said Brooks. “This will help us identify areas for research, as well as better understand the long-term historical statistics.”

 The paper presents a new method to rate the skill of a tornado warning based on the difficulty of the environment. It then evaluates thousands of tornadoes and associated warnings over the continental United States between 2003 and 2017.

The NOAA-funded study finds that nighttime tornadoes have a lower probability of detection and a higher false-alarm rate than the environmental conditions would suggest. Summertime tornadoes, occurring in June, July or August, also are more likely to evade warning.

“The forecasting community is not just looking at the big, photogenic situations that will crop up in the Great Plains,” said Anderson-Frey, the lead author. “We’re looking at tornadoes in regions where vulnerability is high, including in regions that don’t normally get tornadoes, where by definition the vulnerability is high.”

The technique could be applied to forecasts of other types of weather as well.

This research began while Anderson-Frey was a postdoctoral researcher at the Cooperative Institute for Mesoscale Meteorological Studies, a partnership between the University of Oklahoma and NOAA.

This story was adapted from a  University of Washington news release.

Share this:

TORUS project expects “groundbreaking” results

After 32 days on the road, more than 9,000 miles, 19 supercell storms and at least eight tornadoes, researchers expect results from the Targeted Observation by Radars and UAS of Supercells, or TORUS project, to be groundbreaking.

From May 15 to June 15, researchers and students on the project deployed a wide-ranging suite of instruments to collect data on supercell thunderstorms across the Great Plains. The project’s main goal is to determine why some supercells create tornadoes and others don’t.

TORUS brought a unique mix of instruments chosen for the science questions being studied.

A photo of researchers launching a weather balloon in front of a storm.
The windsonde team prepares for a balloon launch during a storm in Nebraska in late May. (Photo by Christiaan Patterson/OU CIMMS/NOAANSSL)

These included mobile Doppler radars and a lidar, mobile sounding systems including a new system that tracks up to eight soundings at once, the NOAA Lockheed WP-3D Orion “hurricane hunter” aircraft, mobile mesonets, and unmanned aircraft systems, or UAS, to sample low-level conditions.

“I am more confident we will make scientific breakthroughs with this project than any other field project in my 16 years of field work,” said Mike Coniglio, a researcher at NOAA’s National Severe Storms Laboratory and a project lead.

Coniglio called gathering the amount of quality data in such a short time impressive.

“It’s not something I would expect we would be able to do, honestly,” he said. “I expected success but we exceeded our expectations.”

Researcher and project lead Erik Rasmussen echoed Coniglio’s sentiments on the project’s success.

“The atmosphere was cooperative,” said Rasmussen. “We have at least four or five cases that will provide the exact type of data we were looking for. Usually, storms are poorly observed, but in TORUS we have at least six storms we collected the sort of data we believe we need to answer our questions.”

Coniglio said TORUS’ success was not just because several tornadoes impacted on the Great Plains between May and June. 

“An active pattern doesn’t guarantee you will get good data,” Coniglio said. “You still have to make good forecasts. We had a better sense of how to forecast these events than we did in the past because convection-allowing model guidance has improved greatly.”

Coniglio said in addition to improved forecasting, the TORUS team’s weather instruments exceeded expectations. UASes launched by the University of Colorado and University of Nebraska-Lincoln performed well. Each UAS had a successful launch, never crashed and received minimal damage from storms.

Rasmussen said the challenge now is combing through the mounds of preliminary data. TORUS acquired more data than expected.

Researchers are currently assembling quality controlled data — basic, quickly compiled data — before in-depth analysis begins over the next four to six years. Rasmussen said preliminary data appears to be intact, with no missing sets, and no instruments appeared to fail in the field.

“When we collect data, we may realize we have something of interest, but we don’t know until

A researcher in a truck preparing equipment before a storm
OU CIMMS Researcher Elizabeth Smith preparing the LiDAR system for operation on the outskirts of a storm. Smith supports NOAA’s National Severe Storms Laboratory. (Photo by Mike Coniglio/NOAA NSSL)

the in-depth analysis,” Coniglio said, who oversaw the operation of a mobile LIght Detection And Ranging, or LiDAR, during the project.

A LiDAR utilizes laser light to detect items like small dust and aerosol particles. Coniglio’s LiDAR team collects observations utilizing the device to track how quickly all the dust, dirt and particles move in the atmosphere.

“The LiDAR saw interesting preliminary differences in airflow among storms and we don’t quite understand that signal yet or what it means, but it is something we will focus on,” he said.

TORUS will collect data again in 2020. Researchers expect to see overarching takeaways based on next year’s data collection.

“This year’s data will help us decide which strategies need to be refined, which tools performed well and if there are any crucial instruments that need to be added,” Rasmussen said.

Share this:

Low Tornado Numbers and Low Tornado Deaths, May 2012-April 2013

Updated 10 May 2013 to add new information from April 2013

Updated 2 May 2013 to correct typo on date of previous low tornado count

The 12-month period from May 2012 to April 2013 was remarkable for the absence of tornado activity and tornado impacts in the United States.

We can start by looking at the number of EF1 and stronger tornadoes during that period. A final count is available through January 2013 and we have a pretty good estimate of how many occurred in February through April, although final numbers won’t be available until July. Although the 12 month total may change a little bit with the final data, it’s unlikely to change enough to affect the results here.

From May 2012-April 2013, the estimate is that there were 197 tornadoes rated EF1 or stronger. Where does that stack up historically? Well, we have pretty good data back to 1954. During that time, the previous low for (E)F1 and stronger tornadoes in a 12 consecutive calendar month period was 247, from June 1991-May 1992. The next lowest (ignoring the overlapping periods, such as April 2012-March 2013) was 270 from November 1986-October 1987. The lowest non-overlapping 12 month counts on record from 1954-present, with the starting month, are:

217 May 2012 (preliminary)
247 June 1991
270 November 1986
289 December 2001
298 June 2000


This apparent record was set less than two years after the record for most EF1+ tornadoes in a 12-month period was set, with 1050 from June 2010-May 2011. The time series showing the evolution of the number of (E)F1+ tornadoes since 1954 is below. The number of (E)F1+ tornadoes in the 12 months beginning with the time on the x-axis is plotted for every month starting in January 1954 and ending in May 2012, the most recent point.

The death toll from May 2012-April 2013 was 7. National Weather Service official statistics go back to January 1950, but we can extend that by using the work of Tom Grazulis from the Tornado Project, who has collected tornado fatality information back into the 17th century. The data are reasonably good back to 1875, but it’s still possible that there are some missed fatalities, particularly as we go back farther in time. So, where does 7 fatalities in 12 consecutive calendar months stack up? Again, here are the lowest totals, going back to 1875, for 12 consecutive months, with the starting month. (For overlapping periods, such as April 2012-March 2013 and May 2012-April 2013, only the lowest period is listed.)

5 September 1899
7 May 2012
8 August 1991
12 November 1909
12 May 1940


Share this:

The Tornado “Drought” of 2012

UPDATE (14-Aug-2012): Graph corrected to indicate 2006 as previous 15 Apr-31 Jul minimum.

The 2012 tornado season in the United States got off to a quick start with well-above average numbers in January, February, and March. Later, over 80 tornadoes occurred on 14 April. Since then, the number of tornadoes in the US has been unusually low. In order to understand how low, we need to look at the long-term history of tornado occurrence. The most reliable portion of the tornado data begins in 1954 but, even after that, we have to be careful in how we interpret it. Since the mid-1950s, the number of tornadoes reported has increased by an average of 14 per year. The increase has been almost entirely in the weakest tornadoes (F0) and is highly likely that the causes are non-meteorological. We can think of this increase in the same way we think of inflation in economics and estimate its impact by adjusting historical tornado counts to account for it. This process, and how it can be applied to part of the year, is discussed here.

That inflation-adjustment process allows us to look at historical data, but a problem still remains of how to look at recent reports. Preliminary, eyewitness reports of tornadoes are collected by local National Weather Service Forecast Offices and the offices then evaluate those reports and produce a list of “final” tornado reports. This process of evaluation takes a few months to complete, so it can be challenging to answer the question “how many tornadoes occurred” shortly after an event. Over the last several years, a simple relationship between the preliminary and final reports has been observed with the number of final reports being approximately 85% of the preliminary reports. As a result, when looking at the preliminary reports in recent months, we can a pretty good estimate of the final reports simply by multiplying the preliminary reports by 0.85.

Let’s look at how many tornadoes we would expect based on the inflation-adjusted tornado count and compare this year’s tornadoes to that long-term expectation. To emphasize the small number of tornadoes since the middle of April, we’ll start on 15 April and add up the number of tornadoes each year through the end of July. In the accompanying chart, we see the distribution of the accumulated number of inflation-adjusted tornadoes as we got from 15 April-31 July. The distribution is based on the period from 1954-2011. The maximum and minimum of any of those years are shown in blue (note that the year associated with the maximum and minimum can change from day to day along the way). The heavy black line is the median of the distribution, the gray lines are the 25th and 75th percentiles (half the years will be between them), and the dashed lines are the 10th and 90th percentiles (4 out of 5 years will be between them). For comparison, the estimated number of final tornado reports from 2012 are shown in red.

Accumulated number of tornadoes from 15 April-31 July from 1954-2011 with 2012 compared to it.
Accumulated number of tornadoes from 15 April-31 July from 1954-2011 with 2012 compared to it.

Through the end of May, the tornado count for the period from 2012 goes along at approximately the 10th percentile of the long-term distribution but, after that, falls well below the previous low. To put this into perspective, the estimated number of final reports from June for 2012 is approximately 100. The previous inflation-adjusted low for any previous June is 94 in 1988. (Remember that the blue line represents the fewest number of tornadoes from any of the 58 years from 1954-2011.) The median number of June tornadoes in 1954-2011 was approximately 270.

July was even more remarkable than June. Only 24 preliminary reports were received, leading to an expected number of final reports of a little over 20. The lowest number of inflation-adjusted tornado reports from 1954-2011 is 73 (1960). Even without inflation adjustment, the fewest number of tornadoes in any July in that time period is 42 (1960), emphasizing the extraordinary nature of this July. The median number of July tornado reports is about 150.

When we look at the whole period from 15 April-31 July, the median tornado count in the record is 850, compared to 2012, with a little under 300. The 850 represents almost 2/3 of the usual annual total of about 1300. One way of thinking about the late spring and early summer tornado season is that the atmosphere missed more than 40% of a typical year’s tornadoes in 3 1/2 months. Compared to 2003, the comparable period in 2012 had more than 900 fewer tornadoes. 2011 had the second highest number of tornadoes in this part of the year, so in the last two years, the US has experienced the extreme high end of the distribution of the number of tornadoes and the extreme low end of the distribution.

Share this:

No Tornado Deaths in May 2012

In May 2012, there were no tornado deaths in the United States. How unusual was that?

We can look at the record of tornado deaths, discussed here, dating back to 1875. The last time there were no deaths in the month of May was in 2005. Prior to that, it was 1994. Overall, there have been 15 years in the 138 years of the record (1875-2012) with no deaths in the month of May, so we’d expect that to happen about once every decade.

May 2012 stands in dramatic contrast to May 2011, when 178 people died in tornadoes, 158 of them in the Joplin, Missouri tornado of 22 May. 178 deaths is the fifth highest death toll in the period 1875-2012, and the largest since 211 people died in 1933. The deadliest May on record was 1896, when 502 people were killed, including 255 in the Saint Louis, MO-East Saint Louis, MO tornado of 27 May. Adjusted for wealth of the country, that tornado was the costliest in US history, with damage adjusted to 2011 dollars of over $6 billion.

Share this:

NSSL scientists study tornadoes in their own backyard

May 24, 2011 Oklahoma tornadoes

Seven destructive tornadoes struck Oklahoma on May 24, 2011.  The tornadoes were well forecast by the National Weather Service (NWS), and NSSL was in position to capture the storms in several ways.

NSSL’s dual-polarized X-band mobile radar captured the early and mature stages of the first tornado reported near Canton Lake, Okla.  The data will be compared with another X-band dual-polarized radar for accuracy.  This storm produced an EF-3 tornado.

The phased array radar successfully sampled a tornadic supercell every one minute as it evolved and went on to produce devastating EF-4 damage in towns west of Oklahoma City, Okla.  A comparison of PAR data with the damage path shows that the radar captured rotation in the storm 12 minutes before it touched down. This tornado was on the ground for two hours with a 75-mile long track.

Visiting forecasters in the NOAA Hazardous Weather Testbed 2011 Spring Experiment found it interesting to be under the threat of tornadoes and then to be in the forecast path of them.  They watched the storms out the window and on the National Weather Radar Testbed Phased Array Radar along with the area Terminal Doppler Weather Radar and the NWS NEXRAD.  These radars showed the evolution of two confirmed tornadic debris balls as both storms moved towards Norman, Okla.  Participants also reported the NSSL/CIMMS weather-adaptive 3D variational data assimilation system (3DVAR) products all handled the track and evolution of the storms and tornadoes very well.

The American Red Cross of Central Oklahoma began using NSSL’s Warning Decision Support System – Integrated Information (WDSS-II) to map rotation tracks of the storm and deploy their teams by 8 a.m. the next day.

And, several NSSL scientists have been in the field as part of NWS teams to survey the tornado tracks and assign EF-Scale ratings based on the damage they find.  The EF-Scale is an estimate of the strength of the tornado based on damage to structures and vegetation.  Preliminary results show three tornadoes out of the seven in central Oklahoma were ranked a violent EF-4.

Share this:

PAR Captures Long-lived Tornado in May 24, 2011 Outbreak

The tornado outbreak forecasted by the NOAA Storm Prediction Center and the National Weather Service Forecast Office in Norman, Oklahoma became a reality as five damage-producing tornadoes struck central Oklahoma between 3 pm and 7 pm CDT May 24, 2011. The longest-track tornado, rated EF3 by the Norman Forecast office, damaged homes and businesses along its 75-mile path that originated just northwest of Binger and moved through the towns of El Reno, Peidmont, and Guthrie (Fig. 1).

Tornado tracks from central Oklahoma tornado outbreak May 24, 2011
Figure 1. Preliminary tornado tracks for the May 24, 2011 tornado outbreak. (Source:

The rapid-scan, S-band phased-array radar (PAR), located within the National Weather Radar Testbed in Norman, Oklahoma, sampled this tornadic supercell every 1 minute. Based on PAR data, by 3:30 pm supercell storm formed its first well-defined hook echo and associated tornado vortex signature about 6 miles west of Binger (TVS; Fig. 2). At this time, PAR data show that the TVS had a maximum gate-to-gate velocity difference of 89 mph. A comparison of PAR velocity data with the damage path shows that the tornado formed about 12 minutes later, at 3:40 pm.

PAR shows strong signs of tornado development. (Image courtesy Pam Heinselman, NSSL)
Figure 2. Based on PAR data, by 3:30 pm supercell storm formed its first well-defined hook echo and associated tornado vortex signature about 6 miles west of Binger.

The 1-minute updates of the PAR exhibit many important details about the evolution of this supercell and its long-lived tornado. One example is the hard-right turn of the TVS and hook at 4:15 pm that placed El Reno in the tornado’s destructive path (Fig. 3 ~62 km northeast of PAR). About 10 min later (4:25 pm, west-side of El Reno), as cells approaching from the southeast began to merge with the hook and a new circulation developed, the hook’s motion was redirected to the northeast, toward Piedmont. Fig. 3 also shows the likely development of two “debris” signatures in the radar reflectivity, which are compact regions of high reflectivity values due to debris from the tornado.

Animated gif of PAR reflectivity and velocity displays
Figure 3. The 1-minute updates of the PAR exhibit many important details about the evolution of this supercell and its long-lived tornado.

This example shows the PAR’s capability to provide timely, detailed information about where a tornadic storm is headed, and its intensity. In the future, this PAR capability may give families the few additional minutes they may need to take cover from destructive storms.

Share this:

10 May 2010 tornado outbreak

(This will be updated as more information and time to post become available. Given the local nature of the event, information may be obtained quickly, but time may be at a premium.)

A significant outbreak of tornadoes occurred over Oklahoma and southern Kansas on 10 May 2010. Numerous damaging, long-track tornadoes have been reported from the Red River on the southern Oklahoma border up through southern Kansas. As of 6 AM CDT, 11 May, there have been 37 preliminary tornado reports. This is likely to change following damage surveys that will begin later today. A very preliminary summary of information from the National Weather Service Forecast Office in Norman covering their region is available here. The Tulsa Forecast Office also has a briefing on the eastern Oklahoma portion of the outbreak.

At this time, there are five fatalities reported in Oklahoma, 2 from the Choctaw area of Oklahoma County and 3 from Tecumseh in Pottawatomie County.

VORTEX2 collected data on the storm that produced the Norman tornado east of Norman.

There were media reports of damage at the National Weather Center. These reports are untrue, although the tornado was visible from the NWC and debris could be seen as the tornado moved east of the NWC. In addition, some staff members suffered damage at their residences.

Share this:

Tornado project enters second data collection phase

Probes I-70
Mobile Mesonets, also known as probes, travel west on I-70 in Kansas.

The Verification of the Origins of Rotation in Tornadoes Experiment  – 2 will begin the second year of data collection on May 1 and run through June 15.  VORTEX2 is the largest tornado research project in history to explore how, when and why tornadoes form.

NOAA and the National Science Foundation are sponsoring more than 100 scientists, students and staff from around the world to collect weather data around and under a supercell thunderstorm.  VORTEX2 teams are using a fleet of 10 mobile radars and 70 other instruments all equipped with cutting edge communication and computer technologies.  Much about tornadoes remains a mystery, and researchers hope this data will help them better understand tornadoes and lead to further improvements in tornado warning skill.

During 2009 operations, the VORTEX2 armada roamed more than 10,000 miles across the southern and central Plains from May 10-June 13.  Data were collected on 11 supercells, including one tornadic supercell.

New for 2010 operations will be the addition of the University of Colorado Tempest Unmanned Aerial System – model airplanes designed to fly underneath the storm to collect data.  Also, three more mobile radars now have dual-polarization capabilities and the radar scouts and mobile mesonets have been redesigned to make operations more efficient.

VORTEX2 2010 operations can be followed on Facebook, Twitter, and through a blog called V2Talk. More information is available on the web:

A special correspondent for kids has joined the VORTEX2 team for 2010.  His name is Chase StormDawg, and he can be followed on Twitter and Facebook too!

Share this:

Some brief notes on the 24 April 2010 long-track tornado

Preliminary information on the long-track tornado that went from eastern Louisiana across much of Mississippi.

1. It had a path length of at least 97 miles. Over the last 40 years, we’ve averaged about one 100 mile path length tornado every 2 years.
Update (27 April): The path length is now given as 149 miles, the 6th longest since 1970. A crude estimate is a once per 8 year event.
2. There were 10 direct fatalities with the tornado. The last double-digit death day was 25 May 2008 (the day that Parkersburg, Iowa was hit.) The last double-digit fatality tornado was on 10 May 2008 (Picher, OK and southwestern Missouri.
3. It’s the most fatalities in a tornado in Mississippi since 21 November 1992 (Brandon-12 fatalities)
Update (27 April):
4. Deaths by circumstance: 6 mobile home, 2 outdoors, 2 vehicle.

Share this: