Lessons From Joplin

Joplin, MO Tornado Through Doppler Radar Storm Relative Velocity

Meteorologists have come a long way in the last several decades being able to provide warnings to the public about severe weather. Doppler radar allows us to detect many tornadoes before they form. On May 22, 2011 the tornado that hit Joplin was warned and the warnings went out as they should have. Still, 159 people died in the tornado making it one of the deadliest in U.S. history. The 1953 Flint, MI tornado was the last tornado to kill at least 100 (that tornado occurred the day before the Worcester, MA tornado that nearly killed 100 as well).

What went wrong? Today the National Weather Service issued their long awaited Service Assessment of the event. One of the biggest problems cited in the service assessment is warning fatigue. Many people heard there was a warning but failed to take action because so many tornado warnings in Joplin didn’t produce a tornado in the city.

The biggest problem with the current warning system is that there is no way to distinguish between a storm that may produce a weak or isolated tornado and a long-track violent tornado that has been on the ground for miles. The current tornado warning program is black and white. Joplin, Tuscaloosa, and Springfield all get lumped in to the same pile as a tornado that sheared the tops off a few trees in someone’s backyard. When there was clear evidence that the Joplin, Tuscaloosa, and Springfield tornadoes were significant or violent tornadoes producing widespread a life-threatening damage there has got to be a better way to differentiate the threat!

Joplin, MO / Courtesy: National Weather Service Springfield, MO

Here are some of the findings from the Service Assessment:

  • Finding #2c Familiarity with severe weather and the perceived frequency of siren activation not only reflect normalization of threat and/or desensitization to sirens and warnings, but they also establish that initial siren activation has lost a degree of credibility for many residents. Credibility is considered to be one of the most valued characteristics for effective risk communication.
  • Finding #2d The majority of surveyed Joplin residents did not take protective action until receiving and processing credible confirmation of the threat- and its magnitude- from a non- routine trigger.

The NWS seems ready to embrace something I’ve been talking about for years. While maintaining part of the existing warning system there needs to be a second tier or warnings for truly extraordinary and life threatening situations. In the last few years the NWS has begun to use the phrase “tornado emergency” to describe strong or violent tornadoes that have either been confirmed on the ground or confirmed by doppler radar (through presence of a debris balls or other means).

There’s no doubt that many people experience warning fatigue when it comes to severe thunderstorm and tornado warnings. A lot of people don’t understand that severe weather can occur in a very isolated area while the area warned is much larger. There has got to be a better way to separate storms that may produce isolated damaging wind gusts or a small, isolated tornado and storms that pose a serious and widespread risk to life.

Taunton, MA Storm Relative Velocity and 0.5 deg. Base Reflectivity on June 1, 2011

The tornado in Massachusetts this year is a good example of a tornado that was an extraordinary event that would be a candidate for a higher level of warning such as a tornado emergency. Unfortunately the Massachusetts tornado was on the ground, inexplicably, for nearly 12 minutes without a tornado warning in effect though there was clear radar evidence that a tornado was likely. In fact the tornado warning that was put in effect after the touchdown was allowed to expire while the tornado was still on the ground! Though there was a serious breakdown in the National Weather Service operations that day for the most part the NWS does a phenomenal job issuing warnings and communicating with the media. Though the Massachusetts case was one of the rare failures the Joplin tornado was an example of a warning issued 17 minutes of lead time prior to the touchdown. Still, even with significant lead time, there has to be a way to reduce the number of fatalities from an event like the Joplin tornado.

Assuming we’re in a perfect world and all tornadoes are detected it is important that we come up with a system to separate the most serious events from events that are more run-of-the-mill. It seems like based on today’s service assessment that may be down the road.

The Hampden County Tornado Through Doppler Radar

In the coming weeks meteorologists will be looking back at the Hampden County tornado trying to learn something from the unique set of meteorological conditions that came together on June 1st. All the signals were there for a classic, significant, and high end severe weather event.

Unfortunately the tornado that hit Springfield did so without warning. The National Weather Service in Taunton, Massachusetts covers Hampden County and didn’t issue a tornado warning until 13 minutes after the tornado was on the ground. It appears part of the issue was that the radar in Taunton was unable to sample the storm well until the storm was over Westfield (and moments away from producing a tornado). The Albany radar, which was initially closer to the storm, had a much better vantage point and had  indications that a tornado may be forming well before the warning was issued.

Here’s the radar from Taunton (BOX) at 4:08 p.m. This image shows reds or winds blowing away from the radar and greens where wind is blow toward the radar.

BOX Storm Relative Velocity 4:08 p.m.

This storm is too far from the radar site and range folding is taking place. RF occurs when radiation from the echo is sent back to the radar site in Taunton after the radar has emitted a second pulse of energy. The radar is unable to resolve whether the energy scattered back to the site is from an echo far away or nearby. The purple shading on the radar is RF.

The Albany radar (ENX), however, is able to sample the echo well without range folding.

ENX Storm Relative Velocity 4:06 p.m.

You can see in this radar shot the tornado vortex signature clear as day 6 miles west of Westfield coming off the Berkshire Hills into the Valley. The “couplet” of greens next to reds indicates winds blowing in opposite directions in a very short horizontal distance. This indicates that at 6,000 feet above the ground this storm is rapidly rotating and capable of producing a tornado. There is about 75 knots of gate-to-gate shear in this image which means the difference between inbound and outbound velocities in neighboring pixels is about 85 mph.

The following radar image at 4:11 p.m. indicates that the couplet has weakened some with only 50 knots of gate to gate shear but 5 minutes later at 4:16 p.m. the couplet rapidly tightens with over 110 knots of gate to gate shear virtually over Westfield.

4:11 p.m. and 4:16 p.m. ENX Storm Relative Velocity

Shortly after these radar scans the storm comes out of range fold on the BOX radar and by 4:22 p.m. BOX has 2 consecutive radar scans with over 100 knots of gate to gate shear between Westfield and West Springfield.

BOX Storm Relative Velocity 4:22 p.m.

The range folding on Doppler radar is unavoidable. However between 4:06 and 4:30 at least one radar (ENX or BOX) indicated a very strong low level mesocyclone during each volume scan. The tornado warning did not come out until 24 minutes after radar indicated a rotating (sometimes very strongly) storm. In fact the tornado touchdown (at 4:17 p.m.) wasn’t even preceded by a severe thunderstorm warning (that came at 4:18 p.m.).

Once the tornado warning came out the radar signatures became more impressive with a couplet, hook echo, and debris ball just as impressive as anything you’d see in Oklahoma. In fact this was probably one the more impressive set of radar images I’ve ever seen.

BOX storm relative velocity and base reflectivity at 5:04 p.m.

This storm shows the limitations and the power of Doppler radar. Switching the radar into VCP 212 with more slices and faster update times increases the range fold problem. Having to monitor multiple radar sites for a storm that can only be sampled between 6-7,000 feet AGL at the lowest slice adds to the difficulty in warning a storm like this. The National Weather Service does a phenomenal job almost all the time with some of the best minds and best technology in the field. Unfortunately this is one of the rare times a strong tornado struck a highly populated area without warning.

Given the amount of YouTube video shot from downtown Westfield of the tornado touching down around 4:17 p.m I am at a loss for how news of the touchdown didn’t reach the NWS. To the best of my knowldge confirmation of the touchdown didn’t make it to the National Weather Service until the storm was over Springfield nearly 15 minutes later! Did local law enforcement or storm spotters not relay information to the NWS?  With a strong tornado traveling through a populated suburb of Springfield in broad daylight it’s a problem that the NWS did not know about it for nearly 13 minutes.

Barnes Airport did report a funnel cloud at 4:24 p.m. but they did not report an actual touchdown. This appears to be the first report of a funnel cloud the NWS received but it’s unclear when the report actually made it to the person who was issuing the warnings.

Being unable to clearly sample a storm near the surface increases the difficulty of issuing accurate warnings. Not getting ground truth reports from affected areas makes the situation even worse.  The National Weather Service also has to be careful not to issue warnings for every storm that rotates to avoid the Boy Who Cried Wolf syndrome.  It’s a fine line that they and the media walk in storms like this.

We will all learn from this storm and hopefully the public will learn how important it is to take tornado warnings seriously.