Friday, March 15, 2019

Tornadoes in New Mexico & Ohio Valley on March 12 & 14, 2019 with record-breaking low pressure system

The so-called "bomb cyclone" that caused hurricane force winds and record low pressure in the Plains on March 13, 2019 also brought tornadoes (see photos above) to southeast New Mexico (NM) on March 12 and to the Ohio Valley on March 14.  The tornadoes at Dexter and near Malaga NM on March 12 were the earliest EF2 tornadoes ever recorded for the state of New Mexico.

Interestingly enough, no tornadoes occurred on the day the surface low was the strongest over Colorado and Kansas (March 13, not shown).  This was because the system was so large and dynamic that it pushed associated surface fronts too far north and eastward, away from moisture and instability.  But that moisture recovered again as surface fronts slowed for severe weather on March 14th over the Ohio Valley.

This system was a good example of how warm fronts are efficient tornado producers. The surface maps below on March 12 & 14th show that the locations of several tornadic storms were near warm fronts: 

Warm fronts offer temperature gradients, increasing moisture, and enhanced low-level wind shear, which is why tornadoes seem to occur so often with storms near such boundaries. (Other tornadoes - not shown - did occur in the warm sector over Alabama during late afternoon and evening on March 14, away from boundaries.)

Note how instability and low-level wind shear combinations using forecasts of the 0-1 km energy-helicity index (EHI) from the RAP model were strong near warm fronts and tornadic storms on both days, offering good support for low-level storm rotation:

The 500mb NAM model forecasts below (roughly 18,000 ft MSL) also show how strong this system was, with a massive full-latitude midlevel trough moving out into the Plains.  A "spreading" jet stream pattern (white arrows) ahead of this large trough generated large-scale lift and ascent over the areas where daytime and evening severe weather occurred on March 12 and 14 (yellow ovals):

Thankfully, no deaths occurred from any of the tornadoes with this system.  But the Colorado blizzard associated with this powerful system claimed at least one life, and flooding in Nebraska from rapid snow melt and heavy rain caused at least one other death.

It is good to see that the week ahead will be much quieter weather-wise across the United States!

- Jon Davies 3/15/19

Monday, March 4, 2019

Many dead from huge Lee County, Alabama tornado 3/3/19

Sunday March 3, 2019 saw the largest death toll in the U.S. from a single tornado since the 2013 Moore, Oklahoma tornado.  As I write this, there are 23 deaths so far from this EF4 intensity tornado in the Beauregard/Marvyn, Alabama area, and more may be found in the next day or two.  (*** UPDATE: As of March 6, all those missing have been accounted for, and the death toll from this tornado stays at 23, with 90 people injured. ***)  With only 10 total deaths in the U.S. last year from tornadoes, the 3/3/19 event is certainly very tragic.

The tornado came from a supercell that entered southwestern Lee County in east-central Alabama around 2:00 pm CST, and the National Weather Service had both a tornado watch and a tornado warning in effect for Lee County when the tornado hit.  But with only 5-8 minutes warning lead time for the Beauregard/Marvyn area, and because this was essentially the first big tornado of a sizable outbreak and was moving fast (around 50 mph!), it probably caught some people off guard.  Also, as is so typical with southeastern U.S. tornadoes, visibility wasn't particularly good (low cloud bases, haze, and trees -- see the photos above and below), which was probably another contributing factor:

Below is a map showing the approximate path of the supercell's tornadic circulation on radar, and also two radar reflectivity images around the time of deaths in the Beauregard/Marvyn area of Lee County.  Note the large classic hook on the inset close-up radar image.

Later images showed a reflectivity "debris ball" (not shown), and correlation coefficient products from radar (not shown) indicated a distinct area of debris aloft as the tornado moved eastward toward the Georgia border.

The killer tornadic supercell occurred in a fairly classic position east of a surface low (see surface map below at 1:00 pm CST), with the storm moving roughly parallel to and just south of a stationary front.  Severe weather forecasters know this to be a rather favorable location for tornadoes.

Combinations of CAPE (instability) and wind shear were also favorable along and south of the front over east-central Alabama at early to mid-afternoon (see significant tornado parameter graphic below  from SPC mesoanalysis at 2:00 pm CST):

Computer model forecasts from the morning of 3/3/19 also did a good job of highlighting parts of central, east-central, and southeast Alabama regarding combinations of CAPE and low-level wind shear that might support tornadoes.  For example, the RAP model forecast of 0-1 km energy-helicity index (EHI) from mid-morning showed sizable EHI values for March near and ahead of the Alabama surface low at 2:00 pm CST (2000 UTC), suggesting good environmental support for supercell tornadoes:

The NAM model forecast of winds and contours at 500 mb (roughly 18,000 ft MSL) for midday on 3/3/19 showed a midlevel trough disturbance aloft (thick red dashed line below) moving toward the southern states.  A "spreading" of the jet stream pattern (white lines/arrows) ahead of this trough provided ascent over the area where the 3/3/19 outbreak took place (yellow oval).  Winds at this level were also forecast to be quite strong, approaching 70 kt, offering good support for supercells near the Alabama surface low.

Lastly, here's a RAP model 1 hour forecast sounding near Columbus, Georgia (Fort Benning, LSF) valid at 2:00 pm CST (2000 UTC), located 25-30 miles east-southeast of the tornadic supercell:

Notice the strong low-level wind shear (around 350 m2/s2 of 0-1 km SRH) and the large looping hodograph (upper left).  Winds at just 1 km above ground (roughly 3000 ft) were 50 kt/58 mph, which suggests impressive wind shear for supporting low-level rotation in storms.  Although the total CAPE wasn't terribly impressive (around 1000 J/kg), notice how much CAPE was centered around 3 km above ground.  This sizable low-level CAPE relatively close to the ground likely helped accelerate air parcels upward in a mature storm supercell updraft, assisting with tilting and stretching of horizontal vorticity from the strong low-level wind shear to support low-level rotation in the Lee County supercell.  

From all this information, sadly, it isn't a surprise that a violent and deadly tornado developed on Sunday March 3, 2019 in this setting.

====>  Consider helping Alabama tornado victims..  Check out options at:

- Jon Davies  3/4/19

Monday, February 25, 2019

First deadly tornado of 2019 in Columbus MS on 2/23/19

Saturday's  EF3 tornado in Columbus, Mississippi (see above) caused the first tornado fatality of 2019 along with several injuries around 5:15 pm CST.  A tornado warning was issued for Columbus roughly 20 minutes prior to the tornado striking town, which likely helped save additional lives.

Farther north, another tornado (EF2, below) hit the west side of Burnsville in the northeast corner of Mississippi, with thankfully no injuries reported:

The setting was well-forecast on SPC outlooks (not shown), and involved an afternoon warm front over central Alabama into northeast Mississippi (see below on College of DuPage surface analysis):
An environment with larger low-level wind shear (0-1 km storm-relative helicity, or SRH, > 300 m2/s2) along and just south of the warm front coincided with sizable CAPE (> 1500 J/kg) over east-central and northeast MS (see HRRR model forecasts below).  This made for a setting supportive of strong supercell tornadoes in the area indicated:

The radar reflectivity image below shows the tornadic cells around 4:40 pm CST, about 35 minutes before Columbus was hit:

As with so many tornadoes in the Dixie states, the Columbus tornadic storm was embedded in a line of storms, which made it hard to see the tornado at times.

The larger-scale setting showed a large trough at 500 mb (roughly 18,000 ft above sea level) plowing through the central U.S.  (A cold-core system was over Kansas, but there wasn't enough CAPE for more than one or two marginal severe reports in western Missouri, not shown.)  Notice the position of the southern branch of the jet stream (thick white line and arrow), with the tornadoes occurring just north or left of that axis, similar to some other tornado episodes I documented on this blog back in early 2018:

The tornadoes also coincided with an area of large significant tornado parameter values (STP, see inset above) over east-central and northeast MS just north of the southern jet branch at 500 mb, similar to the HRRR forecast of enhanced SRH and CAPE shown earlier.

One other point of interest... Tracks/swaths of model forecast storm rotation (called updraft helicity) are a sophisticated tool from the detailed HRRR model used by many forecasters to anticipate supercell likelihood, and by association, possible tornadoes.  Shown below are the accumulated updraft helicity tracks forecast during the day on 2/23/19 by the HRRR model from the ESRL site at mid-morning, the left image showing forecast rotation tracks in the 0-2 km layer above ground, and the right image showing the same for the 2-5 km layer higher up:

Note that the lowest level (0-2 km) updraft helicity tracks are fewer, and appeared to roughly "forecast" the Columbus and Burnsville tornadic supercell tracks (red ellipses).  Although I'm not a fan of using updraft helicity forecasts as a surrogate for the possibility of tornadoes (such forecasts need to be combined with a careful assessment of environment), it is interesting that this case suggests that the lowest level 0-2 km layer product may in a loose sense be more useful in "picking out" some likelihood of tornadoes by reducing the "over-forecasting" of rotation tracks often seen in the 2-5 km layer product.  Just a subjective observation.

- Jon Davies  2/25/19

Tuesday, February 12, 2019

ChaserCon 2019 Forecasting Class slides available online, & "Surviving the Storm" on YouTube!

 Image courtesy of Bill Hark, from Facebook.

ChaserCon 2019 in Wichita, Kansas this past weekend was well-attended, and Shawna and I had a great time talking to so many storm chasers and severe weather enthusiasts!  Thank you to all the convention attendees who stuck around on Sunday, Feb. 10 for my class, and thanks for all the great questions, supportive comments, and big enthusiasm about learning how to forecast!

For those interested, I put up my Powerpoint slides from the class at:

I've also had a surprising number of requests for a DVD presentation on tornado forecasting, similar to my class.  That is definitely something I'll work on in the near future, so stay tuned.  But also be patient, as I have quite a few committments going right now.


One other subject...  My wife Shawna's excellent 30 minute video "Surviving the Storm: What Chasers Want You to Know" is now on YouTube FREE of charge:

This exciting presentation details everything your loved ones and friends need to know about staying safe from a wide variety of severe weather threats.

We can "Pay Forward" our passion about storms (as Shawna has done with this outstanding video) by sharing our knowledge about staying safe when severe weather threatens through talks to community groups, neighborhood home owners and apartment associations, church groups, schools, etc.  This video (no cost!) is a great tool around which to build a severe weather safety talk, if you have an interest in doing that sort of thing.

So please use it and spread the word!  Thanks...

- Jon Davies  2/12/19

Monday, January 28, 2019

Tornado Forecasting class at Chaser Con in Wichita on Feb. 10th !

I'm teaching a 3-hour tornado forecasting class at Chaser Con 2019 in Wichita, Kansas on Sunday, February 10.  It's the first class I'm doing at Chaser Con in 5 years, so I've been working hard to make it interesting and useful for both newer chasers and more experienced ones.

Thanks to Roger Hill for footing the cost and making it part of the convention fee. In the past it has often been an add-on extra cost for convention goers, instead of part of the convention ticket.  So this is a good year to attend.  And thanks to my wife for her patience and helpful comments while I've been putting this together.

I'll do the class in three parts starting at 9 a.m., lasting until around noon with a couple breaks:

Part 1 - General forecasting issues, notes on basic meteorology related to severe weather, and factors more specifically related to forecasting supercell tornadoes.


Part 2 - A detailed case example from 2018 using the material in Part 1, followed by two forecasting exercises for the class.


Part 3 - More advanced topics, including skew T's and hodographs.  I'll also touch on cold-core tornado settings, upslope tornado settings, picking out a cell to follow, and patterns that produce tornadoes on multiple successive days.

While the class won't be streamed online, I've had several people contact me about whether the class will be available later online or by DVD.  I don't know about those issues yet, but my wife Shawna and I will provide some handouts for taking notes, and I'll see about putting all or part of my PowerPoint presentation online for a period after the convention.

There's still time to register for Chaser Con... go to for more information, and click on "Register Now" at the top to register.  Hope to see you there!

- Jon Davies  1/28/19

Friday, December 7, 2018

A December (!) cold-core tornado outbreak in Illinois: 12/1/18

Last Saturday's 28 tornadoes in Illinois (IL) were an unusual event so far north in December. I've had several requests to write about it, particularly because it was associated with a cold-core midlevel low (see this 2006 paper, and this one from 2004).  Here's a diagram showing a common "cold-core" type setting that can produce tornadoes:

As is typical in such settings, the storms on 12/1/18 were low-topped and photogenic, as were the tornadoes.  The images at the top of this post by Jed Christoph show the low-topped supercell near sunset that produced an EF3 tornado at Taylorville IL (22 injuries) at dark.  And below are tornadoes near Beardstown and Havana IL, respectively, from an earlier supercell in west-central IL.  In the first image below, notice the surrounding "blue-sky" appearance so common in cold-core type tornado situations. 
The tornadoes came primarily from these two long-track supercells beginning at mid-afternoon (see satellite photo below, and warning summary graphic from NWS Lincoln IL), and lasted into early evening after dark.  Thankfully, there were no deaths due to excellent warnings and generally good visibility.

Here's the 2100 UTC (3 pm CST) surface analysis around the time tornadoes began over west-central IL:

I've been asked why this outbreak wasn't better forecast.  Actually, SPC did have a well-placed 5% probability of tornadoes over west-central and central IL on their Day 1 outlooks during 12/1/18.  But tornado events featuring strong tornadoes associated with midlevel cold-core lows are, quite frankly, notoriously hard to pick out (see also 7/19/18 in Iowa).

A significant issue is usually instability (CAPE) that appears less robust with a narrower axis than most strong tornado settings we're used to seeing in the Plains.  Because of colder air aloft associated with midlevel lows, what appears to be relatively weak total CAPE can be misleading, with the relevant CAPE usually located closer to the ground in cold-core scenarios. This can set the stage for increased tilting and stretching of vorticity as air in storm updrafts accelerates upward.

Forecast models on December 1 actually did a good job of placing synoptic features and convective precipitation (below).  The 0-1 km energy-helicity index (EHI) combining CAPE and low-level shear often doesn't work well in cold-core tornado settings due to the smaller total CAPE values.  However, in this case, the morning RAP forecast graphics for mid-afternoon on 12/1/18 did suggest that sizable 0-1 km EHI values would be co-located with storms over west-central and central Illinois:

The midlevel low in this case was located farther west than in most cold-core tornado events, 350 to 400 miles west of the surface warm sector and boundary intersection, instead of within 200-250 miles.  But because the low was embedded within a large upper long wave trough in December (see the NAM 700 mb forecast below, first panel), cold air aloft in midlevels had spread far to the east over most of Illinois (note the blue-dotted freezing line in the same NAM forecast panel):

With 50's F surface dew points in the surface warm sector over Illinois beneath the cold air aloft (see surface map earlier), total CAPE values of 1000-1200 J/kg were forecast over Illinois (see second panel above).  This is is pretty decent instability for a cold-core type event, certainly larger than the 200-500 J/kg found in many cold-core settings.

As mentioned earlier, low-level CAPE (in roughly the lowest 3 km above ground) appears to be a significant factor in stronger cold-core events, enhancing low-level stretching.  Typical 0-3 km MLCAPE associated with tornado events is on the order of 50-75 J/kg.  But on 12/1/18, an axis of 0-3 km MLCAPE values near 200 J/kg (see below) was evident over western and central IL on the SPC mesoanalysis at mid and late afternoon, co-located with ongoing convective storms.  Sizable low-level shear (0-1 km storm-relative helicity/SRH of 200-400 m2/s2) was also present:

Such large low-level CAPE isn't directly evident when examining just total CAPE values, and when combined with sizable low-level shear, it can result in strong tilting and stretching of horizontal vorticity in storm updrafts.

I've been experimenting with a version of EHI (0-1 km "enhanced" EHI) that incorporates low-level CAPE to boost the parameter values in situations where large 0-3 km CAPE is present.  Below are both the effective signifcant tornado parameter (STP) fields and enhanced EHI fields for 2100 UTC and 2300 UTC (mid to late afternoon) from the SPC mesoanalysis on 12/1/18.  Note how the enhanced EHI values (EEHI) were larger due to the incorporation of the low-level CAPE in this "cold-core" scenario:

The RAP model 1-hour forecast sounding at Taylorville IL (TAZ) shortly before the EF-3 tornado confirmed the presence of large low-level CAPE (near 200 J/kg), along with large low-level shear (0-1 km SRH around 375 m2/s/):

It's interesting to note that the low-level hodograph on this profile through 3 km above ground looks like an almost "perfect" half circle.  That's very optimal shear available for tilting and stretching into an updraft boosted by large low-level CAPE and upward acceleration in the lowest 3 km.  Having > 60 kts of deep-layer (0-6 km) shear in the environment also helped to organize and strengthen the Taylorville storm's updraft.

Here's an image of the wedge-like Taylorville tornado when it was well southwest of the town before dark:

And here's the tornado moving through Taylorville, at dark, illuminated by power flashes:

Some final thoughts to wrap up here... First, in addition to the seasonal lateness, number/intensity of tornadoes, and westward distance of the midlevel low,  this cold-core event was also somewhat unusual in that the bulk of the tornadoes (particularly the Taylorville storm) occurred well within the warm sector south of the warm front, instead of close to the boundary intersection of the Pacific cool front/dryline and warm front.  That seems to be because low-level shear was strong deep into the warm sector and co-located with large warm sector low-level CAPE.  Cold-core setups in the warm season often lack low-level wind shear except near the warm front and boundary intersection, but that certainly wasn't the case here with this cool season event.

And lastly, this event, along with the 7/19/18 cold core tornado event in Iowa this year, makes me wonder if model forecast graphics of 0-3 km CAPE (not found on commonly-used model forecast web sites apart from the SPC mesoanalysis) should be more available for web users.  An incorporation of low-level CAPE into composite tornado parameters might also help in picking out these stronger cold-core type tornado outbreaks.  

In any event (pun intended), this was certainly a fascinating case for the first day of December.

- Jon Davies  12/7/18  

Thursday, September 20, 2018

Florence spawns a killer tornado near Richmond, Virginia on 9/17/18

Anyone following the news the past week knows that Florence was a devastating hurricane in North Carolina (NC), South Carolina (SC), and Virginia (VA), with at least 37 deaths so far.  The Carolinas deaths were mainly flood-related due to Florence's slow movement and deluge of rain, and with such a large and wet tropical system, tornadoes were probably the least of most people's worries.  But tornadoes did occur, particularly on Monday the 17th when one person died in a tornado that hit the Richmond, Virginia area (see above).

There were around 100 tornado warnings issued during Florence and her remnant's slow journey inland from Thursday 9/13/18 through Monday 9/17/18.  Yet the 17th turned out to be the only truly significant tornado day of Florence's 5-day assault on the mid-Atlantic states.  After my last post about tornadoes from the remnants of Gordon on September 8, I thought it would be interesting to look at possible reasons why the strongest and longest track tornadoes were on the 17th, several days after Florence's landfall.

The inland track of the center of Florence and her remnants from Friday 9/15/18 through Monday evening 9/17/15 is shown below, with times marked:
A few reports of brief weak tornadoes began coming in on the evening of the 15th near Wilmington NC, but the most tornadoes associated with Florence occurred on the 16th (brief and weak), and on the 17th (stronger and longer-lived), as indicated on the graphic above.  But again, why was Monday the 17th the most prolific tornado day?

First, from classic research on hurricane tornadoes, such as McCaul's August 1991 paper, the strongest and most numerous tornadoes with tropical systems are most likely in their right front quadrant (looking downwind along the direction of movement).  More recent studies, such as Verbout et al. (2006) also suggest that tropical systems recurving to the northeast inland over the eastern U.S. are more likely to produce tornadoes.

Such recurvature is usually due to tropical systems meeting and merging with westerly or southwesterly flow from a mid or upper level trough as the remnants move northward (this can increase the surrounding deep-layer wind shear to better support supercells and possible tornadoes -- more on than later).  It is interesting that Florence's remnants did indeed merge with such a trough, as seen on the 500mb charts below (roughly 18,000 ft above sea level) for both 9/16/18 and 9/17/18:

From our tracking chart earlier, notice that Florence's center by daytime on the 17th was moving faster and actually curving back to the northeast, a result of it merging with the westerly flow aloft and the 500mb trough.  Even though Florence's remnants were more spread out at this point, this change in movement and direction still put Virginia in the right front quadrant of the remaining tropical system center on Monday the 17th, a favored location for tornadoes from the research noted earlier.  It is notable that the Virginia tornadoes occurred during the period of Florence's remnants recurving northeastward on the 17th.

Second, looking at tornado forecasting parameters from the Storm Prediction Center (SPC) mesoanalysis, the environment associated with Florence's remnants over land appeared most favorable over Virginia on the afternoon of the 17th, particularly when compared to the 16th.  This can be seen from the Enhanced Energy-Helicity Index (EEHI) parameter, which combines low-level wind shear and instability (important supportive factors for supercell tornadoes), shown below:
Note that EEHI values most supportive of supercell tornadoes remained largely over water on the 16th.  But, on the afternoon of the 17th, large EEHI values came together inland over Virginia where thunderstorms were occurring with the spread out remnants of Florence as the remaining circulation center over West Virginia now moved northeastward.

Although low-level shear near ground (typically in the 0-1 km layer) is a key factor in hurricane tornadoes, deeper-layer shear (throughout the 0-6 km layer) also appears to be a factor for tornado development and strength with tropical systems over land (see my 2006 paper here).  Stronger deep-layer 0-6 km shear (larger than 30 kt or 15 m/s) seems to be important in such cases.  Comparing analyses of this deeper-layer shear (below) on the 16th with the 17th, notice how the 0-6 km deep-layer shear was rather weak (< 30 kt) over the Carolinas on the 16th, but was stronger (> 30 kt)  over Virginia on the 17th where the killer tornado occurred near Richmond. This was a result of Florence merging with the 500mb trough and westerly flow aloft, discussed earlier:
Stronger deep-layer shear helps strengthen and organize convective updrafts, and can help to support tornadoes when combinations of low-level shear and instability are also in place.

It is interesting that on the 16th, with weaker deep-layer 0-6 km wind shear, a supercell tornado over water (a 'tornadic' waterspout) came directly ashore in Myrtle Beach SC (see image below), but produced little reported damage over land (probably EF0 in intensity):

This is in contrast to the killer EF2 tornado near Richmond VA on the 17th (below) when low-level shear and instability combinations appeared more favorable (see the EEHI graphic earlier), and were  supported by an area of stronger deep-layer 0-6 km shear over Virginia (see the 0-6 km shear graphic earlier); this stronger deep-layer shear was absent over the Carolinas the day before.  Here's some more images of the large Richmond tornado:

Distinguishing tropical systems that produce stronger or more numerous tornadoes from those that don't produce tornadoes or are associated with brief weak tornadoes is difficult and certainly not that well understood.   But some of the factors discussed above can be helpful at times in forecasting such tornadoes.

If you can, please donate to a charitable organization to help with recovery from Hurricane Florence.

-  Jon Davies  9/20/18