12 Nov

There Is More Than One Way to Start a Tornado

Of the 2011 Southern tornado outbreak,  April 27th particularly stands out. A total of 199 tornadoes occurred in a 24-hour period leading to 316 fatalities.  That not a misprint.  199 tornadoes.  What makes the event meteorologically interesting is that tornadoes came from  three rounds of weather, each with unique characteristics. Any good, Southern, armchair meteorologist knows the basics of a supercell leading to a tornado. But tornadoes can originate also from quasi-linear convective systems, a only partially understood and complex process. The early morning and midday tornadoes of April 27th arose from just such systems.

From Knup et al. 2014

From Knup et al. 2014.  QLCS vs. supercells in Alabama on April 27th, 2011

When thunderstorms become organized and active at larger scale, the overall complex is referred to as a mesocscale convective system (MCS). When these MCSs approximate something near linear, typically at the leading edge of a cold front, they are referred to as a quasi-linear convective system (QLCS). You may know this better as a squall line. As many a Southerner knows, the squall line contains heavy rains, hail, frequent lighting, and strong winds. Basically your typical Southern spring day.


QLCS over Arkansas

A QLCS can also produce a tornado.  The whole process is very complex and only partially understood.  The cold front lifts the warm air ahead of it forcibly forming the rain line. The rain cools the air causing the air to sink, called a cold pool, which produces strong winds. These winds rushing out causing the squall line to bow.

Ahead of the storm the cold and dense winds force the warmer air to loft. As these winds “empty” the space behind the bow, a low-pressure area is created.  This low-pressure area is filled in by drier air above the storm. This movement continues to accelerates the whole process.  A rear-inflow jet (RIJ) forms caused by the  elevated area of low pressure caused by a tilted updraft over top of the cold  pool.



The bookend circulation at the tips of the bow echo is caused by differences air density due to temperature and pressure, i.e. cold dense air sinks and warm light air rises.  This vertical air movement causes horizontal rotation. Imagine taking a pool noodle, turn it on its axis, and bend it into an arch.



The circulation occurring at the northern part of the bow is amplified due to Coriolis effects.  Interestingly despite the significant rotation,  not all QLCS tornadoes are produced in bookend vortices.  In fact, most  form in smaller-scale vortices at the leading edge of a QLCS.  

bowfujitIn the South a significant number of tornadoes can develop from QLCSs. In one study nearly 55% of the tornadoes in Mississippi and Tennessee over a 5-year period developed from QLCSs.  QLCS tornadoes are unlike supercell tornadoes.  They both form and dissipate quickly and initiate below the radar detection heights. This combinations of factors make it difficult to warn people of a QLCS tornado. Typically by the time the warning is issued… the tornado is already gone. The fact that more QLCS tornadoes occur during the late night/early morning hours make this lack of warning even more concerning.

Screen Shot 2015-11-11 at 6.03.30 PMLuckily, QLCSs do not often form larger tornadoes. Rarely, however, QLCS spun tornadoes can reach EF2-3. This is exactly what happened on the April 27, 2011 (see figure above from Knupp et al 2014). The first 76 tornadoes of the day developed from a strong MCS that developed in Arkansas, grew stronger in Mississippi, and evolved into a QLCS in Alabama in the early morning. In the mid-day a second QLCS developed, producing 7 weak tornadoes. The earlier QLCS tornadoes caused multiple local power outages across Alabama. This reduced the possibility of warnings, as electricity is needed for sirens, radio, and television. The afternoon saw the development of supercells that spawned the largest tornadoes of the April 27th outbreak. Many people never received the warning.



11 Nov

A visit to SWIRLL

Sign outside the tornado shelter or break room depending on the day.

I’m am very late writing this post.  The struggles of being an academic who is working on a book are real.  I was very lucky to be hosted by tornado scientists Anthony Lyza (@tlyzawx), Kevin Knupp, and Ryan Wade (@ryanwadewx) at the University of Alabama, Huntsville back in August.   The new Severe Weather Institute and Radar & Lightning Laboratory, brilliantly named SWIRLL, is fantastic and beautiful facility.  Tony, whose work I’ve discussed previously, took time to show me around SWIRLL.  The three scientists then were amazingly patient explaining meteorological concepts to this neophyte. Overall an amazing day discussing QLCS produced tornadoes (post coming soon), why Southern tornadoes are different, and how topography and vegetation affect tornadoes.  

11 Oct

The 2011 Southern Tornado Outbreak In Numbers

The tornado of April 27, 2011 hit Smithville, Mississippi, where it killed at least 14 people, and moved northeast nearly 3 miles toward the Alabama border. This NASA MODIS image shows the path of exposed ground left in the tornado’s wake. The trail left by the EF5 tornado in Mississippi is much shorter than a similar trail that cuts across northwestern Alabama. Credit: NASA MODIS Rapid Response Team, Jeff Schmaltz

The tornado of April 27, 2011 hit Smithville, Mississippi, where it killed at least 14 people, and moved northeast nearly 3 miles toward the Alabama border. This NASA MODIS image shows the path of exposed ground left in the tornado’s wake. The trail left by the EF5 tornado in Mississippi is much shorter than a similar trail that cuts across northwestern Alabama. Credit: NASA MODIS Rapid Response Team, Jeff Schmaltz

I’m driving along the two-lane highway 40 in Northeast Alabama. A few of hours earlier Tony Lyza and Ryan Wade, tornado researchers at the University Alabama, Huntsville, are giving me directions. I had asked him if you could still see the tracks from 2011 tornadoes. He then begins to give me directions through rural Alabama including a warning about the area known as Meth Mountain. The directions end with statement about watching out for the meatpacking sign.   Sure enough, I see the sign advertising meatpacking services on my right as I am driving along the small highway through Alabama forest. The forest stops and I see nothing but cleared lands with cattle and farms. I say to my wife, “I though this was where it was supposed to be.” All I see is the typical agriculture that dots the Southern landscape. As I look closer the houses are all newer than just a mile back and the few trees both smaller and scragglier. Then it hits me, I am in the damage swath of EF5 tornado. Four and half years later the landscape still bears the witness of the 2011 tornado super outbreak.

Read More

29 Jun

The most unappreciated factor behind Atlanta’s economic growth

Nice write up by David Allison about how air conditioning has fueled Atlanta’s economic growth.

Ask anyone what has driven Atlanta’s economic growth in recent decades and they’re likely to answer the airport, great quality of life, or business-friendly environment. Others might say a strong workforce, the confluence of interstate highways, or great universities. While these are all important, there’s one other key factor that almost nobody thinks about. And that’s air conditioning. It’s at this time of year that this truth becomes most apparent. HOME OF THE DAY SPONSOR LISTING Perfect for Entertaining! See All Homes of the Day Hotlanta’s economic growth, indeed the economic growth of the larger Southeast, almost exactly parallels the growth of air conditioning. Both took off after World War II.

Source: The most unappreciated factor behind Atlanta’s economic growth – Atlanta Business Chronicle

18 Jun

Radar Love


On a Friday in July of 1986, tornadoes touched down in the suburbs of the Twin Cities in Minnesota. A KARE News11 helicopter caught one of these tornadoes on video, a rarity in the era before YouTube videos and hundreds of storm chasers. This video proved to be exceptional, capturing a vortex structure only previously seen in a laboratory setting years earlier by Purdue scientists. When the scientists tweaked the winds entering the bottom and leaving the top of the small laboratory tornado, the twister took on a helical structure. The KARE video was the best documentation to date of an actual tornado with a helical structure, a video that a seven-year-old Robin Tanamachi watched repeatedly. Robin’s father had recorded the KARE broadcast—a videotape Tanamachi quickly wore out from multiple viewings. From an early age, it was quite clear Tanamachi was destined for a career in tornado science.

Almost thirty years later, Tanamachi, now a Research Scientist for the Cooperative Institute for Mesoscale Meteorological Studies in Norman, Oklahoma, examines how radar can be better used to understand, predict, and track tornadoes. Prior to radar, most research on tornadoes was based on films, still photographs, or damage markings (think of the Fujita scale). Although radar originated in the 1950’s, visible radar evidence of a tornado, the tornado vortex signature, was not discovered until 1973 based on observation of a tornado in Union City, Oklahoma. Interestingly, this was the same year that Gold Earring had the hit Radar Love. That discovery by scientists at the National Severe Storms Laboratory in Oklahoma led to the modern tornado warning system in the U.S, including a national network of next-generation Doppler radars (NEXRAD, also known as WSR-88D) funded by Congress.


The velocity couplet

Radars are a paradox of simplicity and complexity. Scientists, like Tanamachi, continue to develop and refine radar technology, the kinds of data radar can produce, and the algorithms used to sort through all this data. At its simplest, radar emits a burst of radio waves moving near the speed of light until they come in contact with a target. The target absorbs and re-radiates the waves back, an echo, to the radar antenna. The larger and/or harder the target (e.g., raindrops or hail), the louder the echo. The time that passes for the wave to leave and echo back is a function of the distance traveled by the wave. Each NEXRAD radar pulse lasts about 0.00000157 seconds (1.57×10-6), with a 0.00099843-second (998.43×10-6) “listening period” for echos. The radar is “on” for a just over 7 seconds each hour. The remaining 59 minutes and 53 seconds are spent listening for any returned signals.

moore_debris_ballThe strength and retrieval time of the echo are just two facets of the radar wave measured. Doppler radar also measures the frequency of the echo waves. A shift in the frequency of the wave can yield an estimate of how fast the target is moving away or toward the radar antenna.   A positive frequency shift implies motion toward the radar and a negative shift suggests motion away from the radar. This is similar to the Doppler shift you hear with sound waves as a loud pick up truck moves toward or away from you.

Because of the earth’s curvature, NEXRAD radar beams typically overshoot the tornado itself, and instead measure the winds in the parent mesocyclone that births the tornado.” The tornado vortex signature appears on the radar as red (indicating high outbound velocity) and green (inbound velocity) pixels occurring adjacent to each other over a relatively small area (see radar image above). This is also called a velocity couplet, and it is associated with the mesocyclone, the rotating vortex of air within in the supercell. Radar can also be used to detect a hook echo extending from the rear part of the storm, resulting from precipitation wrapping around the backside of the rotating updraft. Lastly and perhaps disconcertingly, radar can detect the debris ball from a tornado. Objects lofted into the air by a tornado reflect radar waves very well.

pg29tornadoThe future of tornado research in part, however, rests on getting the radar closer to the tornado. Moving the radar closer to the storm allows the parts of tornado and storm structure to be better observed. Tanamachi is part of team that drives mobile Doppler radars toward, not away from tornadoes, striving to accomplish this daunting task. This is high-risk-but-high-reward science. Out of 100’s of trips, Tanamachi’s team has only successfully collected data a handful of times. “In the unsuccessful cases an ingredient was missing. We will go out if three of the four are present, but even when all four are present something will still be off and a tornado will not form.” But those few successful cases hold one key to the million-dollar question, understanding when tornadoes will form.


Hook echo with potential debris ball

If one key is getting closer to tornadoes with radar, then another is dual-polarized (“dual-pol” for short) radar. Dual-pol radar can emit both horizontal and vertical pulses, allowing for the shapes and sizes of objects to be detected. Contrary to the tear-shaped raindrops of cartoons, raindrops in real life can take on a variety of shapes. Small raindrops are almost perfectly spherical, returning the same amount of echo in the horizontal and vertical polarized channels. Larger drops, deformed by air drag, resemble hamburger buns returning a stronger horizontal echo. Tanamachi notes, “These subtle differences in the drop shape can lead to vastly different estimates of how much liquid water actually falls on the ground. Dual-polarized radar has been conclusively shown to make radar-based rainfall estimates more accurate.”

dual_pol2130425_dual_pol_illustrationThe size a raindrop can also inform meteorologists about cooling and temperatures— possible triggers for tornadoes. Water molecules on smaller drops with their spherical shape and high surface area-to-volume ratio can evaporate more quickly than large drops. This increased evaporation with smaller drops causes the air to chill and sink toward the ground and important in birth of tornado. As Tanamachi explains “Once the cool downdraft reaches the ground, it spreads outward like a puddle of syrup, in what is termed the storm’s “cold pool”. Such downdrafts appears to be a two-edged sword in relation to tornadogenesis. On one hand, several studies suggest that a downdraft is necessary to generate near-surface rotation. On the other hand, the fuel source for the storm (and tornado) is warm, moist, buoyant air, which may be disrupted or cut off if the resulting cold pool completely undercuts the storm, as many a frustrated storm chaser can tell you. It seems there is a “Goldilocks problem” at hand – the downdraft must be present, but not too strong – in order to enhance tornado formation.”  Dual-pol can indicate these inner storm dynamics all by detecting small or large raindrops that in turn suggest the presence and strength of downdrafts leading to predictions on potential of that storm to produce a tornado.

13 May

The complications of measuring tornado strength

Joplin, Mo., June 4 - Tornado Hurls Large Objects –

Featured image from Kansas City District U.S. Army Corps of Engineers (Flickr CC). Strewn debris from the EF-5 tornado that struck the Joplin, Mo., area, shown June 14, 2011.

By the time I was in second grade, I was a professional at school tornado drills. Our school was an old building with a line of nearly floor to ceiling windows lining the entire length of the classroom. Not exactly the best room to wait out a tornado. When 5 short burst rings sounded, I masterly made my way to the hallway with a textbook. Because of its thickness and robust hard covers, I always chose my math textbook.   There crouched down on my knees and elbows, I protected the back of my head and neck with that open textbook. At time I was mastering drills, I was becoming an expert on the F-scale. Indeed, most Southerners know the F-scale as well as they know the words to Sweet Home Alabama. May be better.

The F-scale began with Tetsuya Fujita, a meteorologist from Japan.   The town he lived in at the end of World War II was the primary target of one H-Bombs the U.S. dropped. Due to cloudy conditions that bomb was dropped on its secondary target–Nagasaki. Fujita’s study of the damage of the nuclear bomb blasts actually led to a major discovery of meteorological phenomenon called downbursts and microbursts. Perhaps surprisingly given his history, Fujita was recruited to the University of Chicago in 1953. Eventually in 1971, Tetsuya Fujita and Allen Pearson released the F-scale for tornadoes and Fujita himself quickly became known as Mr. Tornado.

428px-Fujita_scale_technical.svgBefore 1971, all tornadoes were treated the same no matter their strength, size, path, or damage caused. After, thanks to Fujita and Pearson, scientists could begin to quantify differences in tornadoes, a key step into actually understanding tornadoes. The scale ranges from F1 to F12, linking together the Beaufort scale of wind strength and Mach scale (yes like jets). A F1 tornado corresponds to a 12 on the Beaufort Scale and a F12 corresponds to Mach 1. Tornadoes are rated just from an F0 (40-72 mph) to F5 (261-318) that corresponds to increasing wind speeds. An F5 is enough to wipe a significant portion of that sweet home in Alabama completely off the map. Indeed, on April 27, 2011 two F5 tornadoes cut huge tracks from Tuscaloosa to Birmingham and Rainsville to Sylvania. The tornadoes were strong enough to toss a 35.8-tonne (78,925 lb) coal car 391 feet and partially pull an underground storm shelter out of the ground. An F6? Mr. Tornado himself viewed a F6 tornado as completely unconceivable.

Bahari Adoyo Follow Tornado Damage  This is a picture of the tornado damage that my hometown of Tuscaloosa, Alabama took on April 27, 2011. These images have not been edited, some have been cropped. All were taken from a moving vehicle.

From Bahari Adoyo on Flickr (cc) Tornado damage to Tuscaloosa, Alabama took on April 27, 2011.

But here’s the kicker in assigning F-scale ratings to tornadoes.

It is practically impossible to measure wind speeds near the ground. Never mind measuring the wind speeds of a tornado over the course of its life from origination to destruction to demise.   The F-scale is actually a measure of the damage a tornado causes. Or as referred to by Bill Paxton in one of my favorite movies, “It’s the Fujita scale. It measures a tornado’s intensity by how much it eats.”   Assumptions are made that link the damage done, specifically to common wood framed home, to the wind speeds on the F-scale. An F0 cause light damage to chimneys, breaking of trees branches, and damage to sign boards. An F5 causes incredible damage with lifting a framed housed lifted off foundations, carrying it a considerable distances, and the disintegrate it. Cars are tossed through the air over 300 feet. Trees are debarked. Steel reinforced concrete isn’t even safe. The last F5 tornado to occur was in Vilonia, Arkansas in April 27th 2014 just minutes from where I attended college two decades prior.

However as noted by Charles A. Doswell, the meteorologist who contributed to the modern conception of the supercell, the F-scale has issues. “The real-world application of the F-scale has always been in terms of damage, not wind speed. Unfortunately, the relationship between the wind speeds and the damage categories has not been tested in any comprehensive way.” There are several issues that aren’t factored into or estimated in the F-scale.

First, differences in building practices can vary among regions and homes. If your cousin Billy, the strictly for cash under the radar carpenter he is, built the new expansion to your house, chances are the home may not fair very well in even the smallest tornado.

Second, your cousin Billy’s grill, old car, and toilet in his yard, may become tornadic missiles damaging nearby residence. Less junk in the area, less to go airborne, and less damage. 

Third, the steadiness and duration of the wind itself can impact damage. If the tornado is a windbag like Billy, those constant winds will cause more damage. Just like Billy.

Fourth, how much wind does it take to make different cars go airborne, damage different types of wood and frames, and different types of architectural styles? Is a 70’s ranch as likely to survive as the new McMansion over the way in Hope Pines? Does Billy’s rusted 73 Ford long bed fly as far as your new Dodge? None of this is quantified. 

Engineers also noted that it did not take 300 mile per hour winds (F5) to demolish a house. As Doswell noted, “In order to do this right, we would have to do some sort of controlled experiment in virtually every conceivable tornado situation! This is a practical impossibility.”

All the uncertainty in the F-scale lead to meeting in Grapevine, Texas of the Fujita Scale Enhancement Project which brought together a brain trust of expert meteorologists and civil engineers. Ultimately a more objective scale was created, the Enhanced Fujita Scale, which was unveiled in 2006 at a meeting of the American Meteorological Society. Just a year later it went into operational use in the U.S.   The EF-scale has more rigorous and standardized measures of damage, adds additional building and vegetation types, accounts of differences in construction quality, and expands degrees of damage.

But here is the second kicker. While the EF-scale is an improvement, it shares many of flaws of its predecessor. And it comes down to a few simple ideas. Many areas are free from structures and populations. EF0-EF1 ratings predominant in unpopulated areas because of the little damage done to structures. The very famous El Reno, Oklahoma tornado, estimated to have one of the largest diameters ever recorded, could never reach scales of over EF3 because it occurred open country. Moreover, tornado records are plagued with EF unknowns because it requires a detailed assessment. It takes a whole crew in the field after a tornado to assess the damage. Damage for EF-scale is also based on the maximum damage observed in a single spot. But as tornado runs it course, wind speeds can vary greatly. Indeed, the highest damage levels of EF4-5 tornadoes occupy less than 10% of the track.

Clearly there is a need for a better scale for tornado size. Doswell points to a committee of the American Society of Civil Engineers, a sort of Supreme Court for tornado strength assignments. Ratings of tornadoes would be submitted to a committee who would evaluate the proposed rating and ensure standards were met. This committee is developing a standards committee to evaluate all possible ways to measure wind speed though radar, direct measurements, course damage, and even tree falls.   “A study done at turn at the 20th century found if you knew speed and pattern of winds, then directions of tree falls could be use to estimate speed.”

14 Apr

The difficulty of predicting tornadoes


Legendary blues singer, Alger “Texas” Alexander was born in Jewett, Texas, north of Houston in 1900. In 1934, Alger composed and sang with his big, deep voice the Frost Texas Tornado Blues. Just four years earlier a F4 tornado ripped through the town of Frost just south of Dallas. Sixty were killed and many more never found. The entire town was leveled and a mass funeral occurred on the porch of the only house left standing.

I was sitting looking: way out across the world

Said the wind had things switching: almost in a twirl

Says I been a good fellow: just good as I can be

Says it’s Lord have mercy: Lord have mercy on meOkeh_Record_Label

Mmm: mmm

Says I been a good fellow: just as good as a man could be

Some lost their baby: was blowing for two three miles around

When they come to their right mind: they come on back to town

Said rooster was crowing cows was lowing: never heard such a noise before

Does it seem like hell was broke out: in this place below 

frost_tornado_photo_12Although we still sing the blues about tornadoes, much has changed since the 1930’s with regard to tornadoes. We have better home construction and much, much better warnings now.  However, knowing exactly when and where  a tornado will occur…well that is a bit more challenging.

Let me clarify.

With ever-increasing precision we can predict, and warn, of conditions that are likely to produce a tornado. The ingredients—wind shear, instability, heat and moisture, and forcing mechanism—are mesoscale phenomena that can be modeled, quantified, and even predicted. Roger Edwards, lead forecaster at the Storm Prediction Center described the intricate process that it takes to predict a thunderstorm and tornado outbreak days to a week in advance. “It takes a combination of skill, luck, and a team of people. The key is trying to predict a phenomenon not just temperature and rain.” Data of different types—wind, temperature, moisture—at scales ranging from the globe to North America to a particular county in Oklahoma must all be integrated. The data itself comes from satellite, aircraft observations, weather balloons, weather stations, and radar. All of this data must be assimilated and ran through multiple modes that would fry your home computer and probably your neighbor’s on top of that. Then the computers spit out an answer. Of course that answer might be garbage. “Let’s not forget the human element to see that models are going awry. You need to conceptualize in three dimensions what is going on in atmosphere. We still analyze charts by hand which causes us to slow down and think with pencils and paper in hand. Our experience is vital. “

But these ingredients form a supercell, not a tornado itself, and not all supercells produce tornadoes.   “The atmosphere has a way of getting the four together in ways with minor differences to either create a large EF5 tornado or a just some rain. We don’t know when and where these ingredients form in just the right way,” states Edwards. Indeed, 70% of the time a tornado warning is issued no tornado actually forms. It’s important to understand this error would be much worse if it were reversed. In other words, if 70% of the time a tornado occurred no tornado warning was issued. That is the major advance I’m talking about. 

Scientists still have little idea what causes tornadogensis particularly those microscale phenomena, those minor winds, temperature, and pressure differences that occur in areas less than a mile to less than football field that trigger the whole shoot and caboodle. Measuring, analyzing, and modelling this is the cutting edge of tornado research.  In the last several weeks, I have been speaking with the biggest names in tornado researchers from Virginia to Oklahoma to Illinois. I asked everyone what is the biggest question in tornado science. Every one of them responded that the holy grail, my word not theirs, is the when, where, and how of which storm will make a tornado.

Charles Doswell is a meteorologist who although did not originate the concept of supercell, that’s thanks to a Brit named Keith Browning, did with Les Lemon improve Browning’s idea that gave us the modern conceptual model of supercells and how they form. That paper from 1979 is one of the most cited papers in meteorology and tornado science. I’ll leave with Charles’s words from our recent conversation. “We are still struggling with the area of tornadogenesis. Apparently you need every single one of the details…and very high confidence in them too.” Perhaps it’s fitting that Doswell and “Texas” Alexander share more in common that a fascination with tornadoes. Doswell is too a lover of the blues, perhaps fitting for a tornado scientist, and used to host a blues radio show.


08 Apr

Are tornados in the South different?

Image of the Joplin Tornado

Image of the Joplin Tornado

For the last several months, I’ve found myself on a quest. One that as a marine biologist I didn’t expect to find myself on, but that has origins long before I decided ocean science was my path. Tornadoes have imprinted on my psyche. I am not alone in this. Last week, I asked are some areas more tornado prone? What I really wanted to know is if some somehow tornadoes are different in the South. Tornadoes seem to be so much more imprinted on our collective consciousness here.

I posed the question of a difference in Southern tornadoes to Kevin Myatt, an Arkansas native, self-proclaimed weather geek, and writer on the weather blog for roanoke.com. “The first thing you need to know about Dixie tornadoes is they don’t look like the Wizard of Oz funnel” Let me explain. Tornadoes in the South do not come to a neat tapering point. Instead they are sloppy but also key is they are hidden. “Take Joplin, you couldn’t even tell it was a tornado except from one angle.” Higher moisture in South due to the proximity to the Gulf of Mexico leads to wetter storms. The tornado becomes rain wrapped. Combine this with a cloud base close to the ground and you have tornado edges that blurry and shrouded.

tumblr_lzg94uCCHT1ql23g2o1_500 giphyBut there is another key idea. Southern tornadoes are the danger that lurks in the night—the proverbial boogey man. Both Myatt and Roger Edwards, lead forecaster at the Storm Prediction Center that I also posed my question to, both mentioned night tornadoes to me. Tornado season in the South includes November to February when days are shorter makes them hard to see. Combine this with the more complex topography and forests of the South over the Midwest and you have a recipe for distress.

tornado-watches-per-year1999-2008 tornadoes_countyThe reason is not because more tornadoes occur in Dixie—Kansas, Oklahoma, Texas, Nebraska, Colorado, and Florida safely hold that title. But interestingly, outside of Oklahoma, more tornado watches occur annually in Arkansas, Louisiana, Mississippi, Alabama, and Georgia. Parts of Mississippi and Alabama see over 16 annually. Edwards states, “In reality, tornado season is different in South because it can be anytime of the year.” During the early part of the year, Gulf of Mexico moist air brings moist, warm, and unstable that will turn a normal thunderstorm into a tornado. In the fall, hurricanes and tropical storm spawning tornadoes. Whereas we may not see more tornadoes, we are definitely living under the constant threat of them.

concentrated-poverty-for-map-exchange-may-2011Edwards of course also points to poverty, flimsier homes, and poor building codes, prevalent throughout the South, often lead to more damage even in weaker tornadoes. “Preparedness and construction are vital.” Apparently the constant threat has made us more complacent rather than more prepared.

02 Apr

Are some areas more tornado prone?

Screen Shot 2015-03-18 at 11.00.04 AMConway rests almost square in the middle of Arkansas. In the early 90’s when I attended college in this Southern town there were fewer than 30,000 people.   Outside of the Old South Pancake House, a Wal-Mart, and the annual Toad Suck Daze, nothing much occurred, not even tornadoes, a rarity in a town also squarely in Tornado Alley. Just a few minutes away sets Vilonia, a town of a few thousand that makes Conway look like a sprawling metropolis. What Vilonia lacks in cultural exhilaration is more than made up for in meteorological excitement. In 2011 an EF2 tornado wiped out a small part of the town. In 2014 and EF4 wiped out the rest. But Vilonia has nothing on Moore, Oklahoma. In 1998, 1999, 2003, 2010, twice in 2013, and again 2015, tornados have struck this town just south of Oklahoma City. Since the town’s founding in the late 1800’s over 20 tornados have removed parts of this small town.

All this begs the question, are some areas more tornado prone? 

Screen Shot 2015-04-01 at 9.54.24 PMOver very broad scales this is most certainly true. The area known as Tornado Alley sees the clash between warm moist air from the Gulf of Mexico near the ground, colder air in the upper atmosphere from the west, and a third layer of very warm dry air between the two levels from the southwest that tries to keep the other two at bay (see background at this post). 

Although it covers just 15% of the U.S., Tornado Alley lays claim to nearly 30% of all the confirmed tornadoes in the Storm Prediction Center’s database between 1950 and 2012. Of the 58,046 tornadoes on record in that period, 16,674 of those occurred in Tornado Alley, which is a long-term average of 268 tornadoes per year.-ustornadoes.com

Screen Shot 2015-04-01 at 9.54.17 PMBut at more local scales are some areas “protected” or “off limits”? Over a decade ago, meteorological researchers Chris Broyles and Casey Crosbie at the Storm Prediction Center in Norman, Oklahoma proposed that are “smaller tornado alleys.”

Many smaller tornado alleys were identified across the Mississippi Valley, Tennessee Valley, Great Plains, Ohio Valley and Carolinas. Though there may undoubtedly be specific meteorological reasons why these apparent alleys exist, one hypothesis is the smaller alleys are related to topographic features that may modulate environmental conditions in ways that favor development of these types of tornadoes.

One adage is that “tornadoes don’t happen in the mountains”. As someone who grew up in the Ozark Mountains and saw my high school demolished it’s not a belief I personally hold but anecdote does not make science. It does appear that tornados are less frequent in the mountainous areas due increasingly colder temperatures with increased elevation. This cold denser air at higher elevations is more stable—not exactly the best trigger for a tornado. Yet, while tornados are less frequent they do still occur in the mountains. Just as one example, an EF3 tornado hit at 2,080 feet over Glade Spring, Virginia in April 2011.

Researchers at the University of Alabama at Huntsville have also found the roughness of that topography can also influence the power of tornado. Kevin Knupp lead of the research team states, “Forested areas have a rougher surface than open agricultural regions. Forested regions over mountains are even rougher because the mountain topography has a certain roughness associated with it.” In simulations, the rougher the area the stronger and wider a tornado can get in simulations. Another researcher in this group, Anthony Lyza, has started providing evidence that tornadoes in Alabama are affected by topography. Tornadoes weaken as they proceed up and strengthen as they proceed down mountains and hills. Sometimes whether uphill or downhill a hill or mountain will just cause a tornado to dissipate. Circulation will intensify as a tornado moves onto and weaken as it moves off a plateau. Tornado tracks will deviate to follow plateau edges and valleys.

Screen Shot 2015-04-01 at 10.23.10 PMIn the map above (from Broyles and Crosbie) you note a mini-tornado alley in the Northeast Arkansas. Changes in topography from the forested mountains of the Ozarks to the flat farmland of the deltas of the Mississippi and Arkansas Rivers lend some credence to Knupp’s hypothesis. In combination with this topographic change, Gulf of Mexico moisture running up the Mississippi river may also bank against the mountains leading to minor atmospheric instabilities that trigger tornadoes.

Yet variance in the number tornadoes from location to location may be just biases in reporting and spotting tornadoes in unpopulated regions. James Elsner and team at Florida State University found that “historically, the number of reported tornadoes across the premiere storm chase region of the central plains is lowest in the countryside,” a pattern greatly mitigated in recent years due to storm chasers.


Top a uniform distribution. Middle a random distribution. Bottom a clumped distribution

There may be another reason for mini-tornado alleys. Randomness. Humans are almost completely useless in detecting spatial patterns. To clarify, we specifically have difficulty separating random from clumped patterns, especially when the clumpiness of events is weak to moderate. Keep in mind that small amount of clumping happens in truly random processes. Take the figure below. I generated 1,000 random numbers between 0 and 1. I generated another set of 1,000 random numbers between 0 and 1. I used these to represent the x,y coordinates of simulated tornado touchdowns on my hypothetical tornado alley. You can clearly see that simulated tornadoes are clumped in some regions. Compare this to a figure James Elsner posted on Twitter (far bottom) of actual tornado frequencies in Kansas.The next step for tornado research is to separate random pattern from actual areas that see more than their fair share of tornadoes.




Tornado paths 1954-2013 over the central Plains. Counts per grid cell.


24 Mar

The Making of a Tornado


A siren goes off in the distance. It’s not stopping and the anxiety I feel rises with every second it continues. That sound, a warning, signals conditions are “right” for a tornado. Even though that siren just started, the ingredients that brought us to this moment were mixed much earlier.

Building a Supercell 

Everything begins in the upper parts of the troposphere, which roughly spans the lowest 6 miles (10 km) of our atmosphere and contains all of the earth’s weather. Thermal winds, as the name implies, were caused by differences in temperature. However, contrary to the name a thermal wind is not really a wind. There’s an old saying, “You can’t fly a kite in thermal wind.” Technically, it’s the difference in wind between two levels of the atmosphere. Colder air is denser than warmer air. This creates differences in thickness values, the vertical distance between two pressure levels. Think of a sandwich under a brick. The differences in the thickness values from cold to warm create a pressure gradient generating the speedier winds

Thermal wind applied B. Geerts and E. Linacre http://www-das.uwyo.edu/~geerts/cwx/notes/chap12/thermal_wind.html

Thermal wind applied B. Geerts and E. Linacre http://www-das.uwyo.edu/~geerts/cwx/notes/chap12/thermal_wind.html

Winds in the upper troposphere are now moving faster than wind closer to the ground. This creates vertical wind shear, more simply put is a change in wind speed or wind direction with height. Much like a paddle wheel, this wind shear generates horizontal rotation. This is the rotation of the tornado in its very infancy but much more is needed for birth. 

National Weather Service http://www.srh.noaa.gov/jetstream/tstorms/windshear.htm

National Weather Service http://www.srh.noaa.gov/jetstream/tstorms/windshear.htm

The second ingredient-rotating updraft

For the next step, this horizontal rotation needs to become vertical. In the middle troposphere, westerly winds are transporting a cool, dry air mass over the warm moist air coming from Gulf of Mexico. The overlap of these two air masses creates instability. The hot air wants to rise because it’s less dense. The energy of the rising air can be measured as convective available potential energy (CAPE), the total amount of energy a parcel of air would have if lifted a certain distance vertically through the atmosphere. Higher CAPE values indicate more energy. This updraft can tilt the horizontal rotation into vertical rotation. 

However, a cap of warmer air prevents this. This layer of air originated in the desert southwest. Some of this Southwestern air gets lofted. The warm dry air eventually rides over the warm moist air from the Gulf of Mexico. Now warm, moist air from the Gulf is beneath the warm, dry air from the desert Southwest. The cap is stable and prevents the updrafts from penetrating very high into the atmosphere, but as the day progresses, the conditions change. By mid to late afternoon, during peak heating, the rising air from the surface layer of air is warmer than the cap. The cap is broken and air can now ascend several miles into the sky. A thunderstorm with a rotating updraft, a supercell, has now developed. 


Supercell.svgTurning a Supercell into a Torndao

Not all supercells produce tornadoes. Indeed, 70% of the time a tornado warning is issued no tornado actually forms.  The when, where, and how of which storm will make tornado is one of the biggest questions in tornado science. Scientists hypothesize the key to tornadogenesis is a downdraft occurring on the backside of the storm, the rear-flank downdraft (RFD). This downdraft forms as some of the falling precipitation wraps around the updraft producing a characteristic hook echo on radar. The RFD brings rotation from aloft to the ground. The RFD can even generate additional horizontal rotation that can be titled into the vertical rotation. This results in a region of broad rotation at the surface, but a tornado is defined as a “violently rotating column of air in the contact with the ground.” As the air from the RFD converges beneath the updraft, it can be ingested by the updraft, amplifying its rotation exponentially. Think of the age-old physics example of a figure skater bringing his or her arms closer to increase the speed of his or her spin. The air is more likely to enter the updraft if it is relatively warm, since warmer air is less dense than cold air. With the third and final ingredient the warning has become reality.


 Special thanks to Gabe Garfield (@WxGabe), research meteorologist for the National Weather Service,  and Jeff Frame (@VORTEXJeff), Assistant Professor of Atmospheric Sciences at University of Illinois, for both teaching me about tornadoes but providing feedback on this post.

Featured image above from Daniel Rodriguez on Flickr.   El Reno EF-5 Tornado taken from just north of Banner Rd and 15th Street.