Wednesday, June 5, 2019

A merger of two separate long-track tornadoes in northeast Kansas on May 28, 2019 !

I wasn't planning on posting anything further about the Lawrence-Linwood tornado and the Silver Lining Tours (SLT) encounter with a separate rain-wrapped tornado, thinking enough has been said already.

However, the National Weather Service in Topeka today released an addendum to their original survey identifying and adding a new tornado (EF2) that was previously unreported in southwest Douglas County, Kansas.  This was the rain-wrapped tornado that struck SLT, rolling two of their vans, and was located at the time of the incident within the rain-filled rear flank downdraft (RFD) of the developing mesocyclone to its north that gave birth to the large Lawrence-Linwood tornado.

As I suggested in my prior post, this additional tornado was not a satellite tornado, but a separate tornado that had not yet been confirmed or reported previously.  It ended up merging with the developing large Lawrence-Linwood tornado, a complex and fascinating evolution and interaction.

Here's a map I constructed by putting together NWS Topeka and NWS Kansas City survey maps of the two tornadoes.  This shows the full tracks (nearly 43 miles total) and where the two tornadoes merged:


Roger Hill was kind enough to share his video with me privately, shot while SLT was retreating to the south away from the developing Lawrence-Linwood mesocyclone and what they thought was away from danger.  With Roger's permission, here's a panoramic image I put together from his video as they were pulling away to move south:

You can clearly see the visual mesocyclone to their west-northwest (at right), also the focus of several other chaser videos online.  But also notice the advancing rain-filled RFD coming up from their south (center of the image) that unbeknownst to them contained a hidden and unreported tornado that no one yet knew about.

Most spotters and chasers are taught to stay southeast of northeastward-moving mesocyclones (areas of organized rotation in supercell storms) to stay safe.  So, SLT thought they were doing the careful thing to head back south to Highway 56 and follow the primary mesocyclone from a safe distance.

Complaints and accusations online (Facebook, Storm Track, etc.) have focused on the question, "What was SLT even doing in the bear's cage?"  Well, that question kind of misses the point when the room becomes dark and you don't even know an additional bear's cage is there after you're already steering clear of the main bear's cage that you know is behind you.

The tornado merger in this case was very similar to the Hesston-Goessel, Kansas tornado merger I studied on March 13, 1990 that led to an EF5 tornado:

The difference is that the 5/28/19 merger took place hidden in rain within an HP storm, while the Hesston-Goessel supercell was a classic non-HP storm and very visible.  In fact, we might not have even known about this merger had SLT not had their encounter!

To my knowledge, this is the first time a merger of two separate medium or long track tornadoes has actually been documented within an HP supercell, although similar mergers have undoubtedly occurred but not been documented.  This case certainly deserves further study from several perspectives.

Thanks to meteorologist and severe storms expert Greg Stumpf for being astute enough to recognize the similarity to the 1990 Hesston-Goessel merger/interaction, and mentioning it on Facebook already several days ago.

- Jon Davies  6/5/19

Sunday, June 2, 2019

Radar data and the Silver Lining Tours incident 5/28/19 south of Lawrence, Kansas

Roger Hill and Silver Lining Tours (SLT) have been getting a bunch of heat on social and other media for two of their tour vans getting rolled during last Tuesday's tornadic storm south of Lawrence (see my post about the weather setting here).

I'm certainly no radar expert, but based on radar data I went through carefully on Saturday, I do think everyone needs to chill a bit before passing judgement about Roger and the tour group being "too close".
 
A discussion on Storm Track forum has been looking at the early stage of the Lawrence tornado's track via Topeka (TWX) radar, and there was definitely some complex evolution going on with the storm.   This data does suggest that the main tornado was developing a couple miles to the north of SLT's position around the time they were struck by strong winds or some sort of 2nd rain-wrapped circulation coming up from the southwest in an unusual location, which matches statements by Roger.

I also looked at radar data saved from the Kansas City radar (EAX, a little farther away), and can make out two separate circulations at 6:00 pm CDT in the storm-relative velocity field:







This was just before the large tornado developed at 6:05 pm CDT, with the 2nd smaller circulation visible closer to Highway 56.  This 2nd circulation was located about 2 miles (check the scale on the graphic) to the south or south-southeast of the developing large tornado.

Base reflectivity at the same time (below) showed the large developing tornado circulation to be within some rain, with the 2nd circulation located southward within the wet rear flank downdraft (RFD) that was beginning to surge east with this large HP supercell:


Discussion on Storm Track has highlighted this additional circulation to the south of the axis along which the large tornado formed, based on radar images and other information.  Here are radial velocity images from TWX (courtesy of Jeff Snyder on Storm Track) at 5:56 pm, 6:00 pm, and 6:04 pm CDT.  I've marked the locations and tracks of the two circulations.  Notice that it appears the southern circulation moved right over the location where the SLT incident occurred, which lends credence to Roger's description of the incident (click on the image to view it larger):






Dan Robinson on Storm Track put together a graphic suggesting the evolution of the two circulations discussed above.  Here it is, a rather complex evolution prior to the start of the large tornado, with a red "X" marking SLT's location:


Quincy Vagell was shooting video at about 6:00 pm CDT from near the location of the SLT incident,
looking toward the north-northwest:

Notice the lowering visible to the northwest (in spite of rain), which is probably the developing circulation and feature that spawned the large tornado at 6:05 pm, matching what Roger and SLT say they were watching from the southeast.

Quincy describes on Storm Track a "bluish" rain curtain that was moving up from the southwest behind him and to his left.  This was probably the wet RFD containing the small embedded circulation closer to Highway 56 that was not visible, and may have been what hit SLT.

To give some scale and context, I'll mention that my wife Shawna and I watched last year's Tescott, Kansas EF3 tornado on May 1, 2018 (a "classic" non-HP supercell) develop from a distance of about 2 miles to our northwest and north:






















This was a similar distance and position to what Roger and SLT had on 5/28 relative to the lowering that they could see to their northwest, which appeared to spawn the large Lawrence tornado a few minutes later.  But, with the wet RFD curtain coming up from the southwest, there was no way to see the 2nd circulation coming with it (from an unusual location within the storm) as they drove back south toward Highway 56 in order to head east and follow the storm from what they thought was a safe distance.

As I mentioned in my prior post, my wife Shawna and I got caught in a somewhat similar surprise situation (strong surging wet RFD winds from the southwest, south of a large tornado) with the El Reno storm in 2013, even though I have years of experience chasing storms as a meteorologist. Luckily, we managed to drive out of the RFD winds without incident.

I think what has sparked accusations of being "too close" in this situation is the characterization of the 2nd / southern circulation shown in the radar images above as a "satellite" tornado, something that implies being really close to a larger tornado.  In fact, this 2nd circulation was separated from the developing main tornadic circulation by at least 2 miles or more, until it moved toward the developing large tornado after striking SLT.  As such, it seems part of a complex evolution that gave birth to the tornado, and not a true "satellite" tornado.

I talked to Roger briefly on Saturday, and know that he is very shaken and more than a little confused by the evolution of what happened last Tuesday.  I think we should all give him the benefit of the doubt on this event.

He and I also talked about swearing off HP storms in the future.... they often surprise and do the dangerously unexpected.

Looking at the bigger picture, maybe spotter and chaser training should start focusing more on large HP storms as a special case with elevated dangers, emphasizing the possibility of wet RFD surges and dangerous winds as an additional hazard south and some distance away from a rain-wrapped tornado.  And, maybe training should teach and emphasize, for safety's sake, that spotters and chasers give large HP storms a much wider berth than what would seem necessary with storms that are more "classic" in nature.  Large tornadic HP supercells are complex and visually challenging, even for experienced spotters and chasers.

Shawna and I are incredibly glad this did not turn out to be another deadly situation, and that everyone with SLT survived.  That’s what's most important.  Just my two cents worth…

- Jon Davies  6/2/19

Thursday, May 30, 2019

A violent tornado close to home: The May 28, 2019 EF4 Lawrence-Linwood, Kansas tornado



The third violent (EF4 intensity) tornado in the U.S. so far this year occurred on Tuesday May 28 near Lawrence, Kansas (KS) and west of Kansas City, Missouri (MO), not far from my home.  It was the first violent tornado to affect the Kansas City metro area since 2003, and sent my wife Shawna and our cats to the basement for the first time ever.

We were on a 3-day Memorial weekend trip to the High Plains of Colorado (our first multiple day chase trip in over 3 years), but awoke Tuesday morning in North Platte, Nebraska to find Kansas CIty under the gun.  So, instead of chasing north-central Kansas, we rushed back to Kansas City to check on family and cats, and found ourselves in our basement taking shelter not long after we got home.

The track map below shows how the tornado would have come close to our house had it not dissipated near Bonner Springs, KS.  We're extremely grateful that the tornado did not directly affect us or family members, but are sad for those near Lawrence and Linwood, KS (see photos of the tornado near those locations above) who lost property and homes.  Good warnings and media coverage undoubtedly saved lives, but 18 people were injured.


I've had several people ask me why the tornado was violent and long-lived (nearly 32 miles and just short of an hour on the ground).  The surface map at 4:00 pm CDT showed a stationary front (reinforced by earlier storms) draped acoss northeast KS and northwest MO, with the Lawrence-Linwood supercell taking shape near Emporia, KS (circled "S"  on map below), in the warm sector well south of the front:



Computer model forecasts from the morning suggested that low-level winds and the low-level jet would strengthen over eastern KS during the day, nearly doubling in speed between noon and 7:00 pm CDT on NAM forecast graphics for 850 mb (roughly 5000 ft MSL):


When we arrived home at late afternoon after passing through the front from north to south, we noticed that winds were strong from the southeast, which tended to confirm the model forecasts.  This in turn increased low-level shear, which combined with instability resulted in increasing energy-helicity index (EHI) values during the afternoon, supportive of significant tornadoes (compare the RAP model EHI forecasts for noon and 5:00 pm CDT below):



Typically, the low-level jet is at a diurnal minimum in speed at late afternoon in late spring. But on May 28, a strong upper system (see the NAM 500 mb upper air forecast in midlevels for 4:00 pm CDT below) was lifting northeastward through the Plains.  The increasing low-level jet and 850 mb flow was in response to this strong dynamic system:


Going back to the surface map earlier, notice the backed southeast winds at Topeka (TOP) and Olathe, KS (IXD).  That further increased the low-level wind shear ahead of the soon-to-be tornadic supercell, and likely helped it to produce the large tornado long before it interacted with the front farther north.

The HRRR model forecast of the fixed-layer significant tornado parameter (STP) also suggested strong support for tornadoes at late afternoon over northeast KS:


A dangerous aspect of this violent tornadic supercell was that it was very HP (high-precipitation) in nature, wrapping the large tornado in rain, making it difficult if not impossible to see.  The views of the tornado at the top of this blog post are from the north or northwest looking toward the south or southeast, where the back edge of the tornado was sometimes visible.  However, from the east, northeast, or southeast, a large rain-wrapping shield hid the tornado, as in this view toward the west or southwest from Highway K-10 east of Lawrence:


The experience Shawna and I had with the El Reno tornadic supercell back in 2013 has forever made me very leary of HP supercells, so much so that I now always put lots of distance between me and such storms.  In fact, I tend to avoid chasing them.  That's because HP supercells are so difficult to view and assess both storm movement and structure, not to mention any tornado, which is truly frightening.

A related dangerous factor of HP storms is the rain-filled rear flank downdraft (RFD) that often surges and wraps around the tornado from the south and southeast.  On the radar reflectivity images below just after 6:00 pm CDT, notice how the wet RFD surges eastward on the south and east side of the tornadic circulation.  If trying to stay with the storm by stair-stepping eastward and northward, it is not unusual to have this surge overtake you, as can be seen along north-south Highway 59 south of Lawrence in the images below:


My experience in the field is that rain-filled RFD surges are often strong wind producers (apart from any nearby tornado), as with the 2013 El Reno tornadic storm where we nearly got blown off the road retreating south of the intensifying and rapidly expanding tornado.

I've seen a couple videos of chasers following the May 28 Lawrence-Linwood tornadic storm way too close, hooping and hollering while surrounded in blinding rain and seemingly oblivious to the danger near them.  I used to do stupid stuff when I was younger 25-30 years ago, but experience and maturity in recent years has taught me to give these storms a very wide berth, even if it sacrifices the chance to view a large tornado.  I fear that more chasers will get killed in future years on HP-type storms of this nature.

It is truly miraculous and awesome that no one got killed with the May 28, 2019 Lawrence-Linwood tornado.  That is something to celebrate.

- Jon Davies 5/30/19

Monday, May 13, 2019

Boundaries aiding or inhibiting tornado potential: Examples on May 5, May 6, and May 7, 2019 in the Plains


A week ago, May 5 - May 7, 2019 saw several supercells and tornadoes over the central Plains (for example, see images above by Tony Laubach on May 6 after dark in central Kansas ).  Thankfully, there were no serious injuries from any of the storms.  But boundaries on these days certainly had some influence on whether particular storms did or did not produce tornadoes.

This post will look at boundary details that likely contributed to certain storms being tornadic or non-tornadic on these three days, using surface, satellite, and radar data.

First, on Sunday May 5, an outflow boundary was leftover from morning convective showers over central and southern Kansas, seen on this noontime visible satellite image:
By mid-afternoon, this boundary was still visible over central Kansas, along with a dryline wind shift boundary over southwest Kansas that was marked by newly-developing cumulus (also see the 4:00 pm CDT surface map below):
By 5:00 to 6:00 pm CDT, storms had developed from Kansas to west Texas:
It's notable that the storm/storm cluster in central Kansas near the old outflow boundary on May 5 produced tornadoes off and on for at least a couple hours as the cluster moved southeast near this boundary, including this EF1 tornado northwest of Hutchinson, Kansas:
The SPC mesoanalysis depiction of the significant tornado parameter (STP) at 5:00 pm CDT suggested that factors supporting supercell tornadoes (such as wind shear and CAPE) were maximized somewhat in central Kansas, probably due in part to this boundary:
It's interesting that the morning HRRR model forecasts (not shown) did not show any storms developing along this boundary.  The tornadic storms on May 5 in central Kansas near Great Bend and Hutchinson probably would not have occurred without the remnants of the morning outflow boundary enhancing convergence and local wind shear.

A landspout also occurred on May 5 northwest of Dodge City between 5:00 and 6:00 pm CDT, as storms developed along the sharp wind shift boundary in southwest Kansas seen on the satellite photos earlier, where surface heating and CAPE intensified along the vorticity-rich boundary.
A large dusty tornado (EF2) also occurred with a supercell in west Texas south of Lubbock (image below), within the warm sector and not associated directly with a boundary.  This was the strongest and longest-lived tornado of the day, and formed ahead of  a dryline bulge within an environment of enhanced STP values in that area (see earlier STP graphic).


The next day (Monday, May 6), the 7:00 pm CDT surface map showed a slow-moving east-west front over central Kansas that initiated storms during the late afternoon and evening:
Between 6:30 and 7:00 pm CDT on May 6, storms on radar were ongoing in central Kansas, including a cell that was tornado-warned near McPherson (MPR, south of Salina/SLN).  Note the blue "fine line" on the radar-image below, located just _south_ of McPherson and indicated by the small white arrows:
This strongly suggests that the tornado-warned storm was undercut by cold air north of the front, which is probably why the storm did not generate a tornado, although it did produce large hail (2"+).  A photo of the McPherson storm also visually suggests it was occurring atop cold surface air, with lines of scud clouds visible near the ground and a laminar shelf forming:
Another tornado-warned storm was located farther west, northeast of Dodge City (DDC), also just north of the surface front (dashed line below) with cold air likely undercutting it, as suggested by the frontal "fine line" position (not shown) from the Dodge City radar:
As the evening went on, the boundary slid southward through Dodge City (still visible as a "fine line"), but then slowed down as storms consolidated into a significant supercell just northeast of Kinsley, Kansas after 9:00 pm CDT:
Notice that the boundary east of Dodge City (as marked by the dashed line above) had gotten "pulled" back north into the strong supercell northeast of Kinsley so that the storm could access the unstable warm sector air mass, instead of being undercut like earlier with cool post-frontal air.  That likely had something to do with the supercell producing a tornado after dark east-northeast of Kinsley shortly before 10:00 pm CDT (see images at top of this post):

The SPC mesoanalysis depiction of STP at 9:00 pm CDT showed an environment supportive of tornadoes feeding into this supercell from south of the boundary:
It's also interesting that, farther west,  a tornado occurred around 9:20 pm CDT at the location where the aforementioned boundary hooked into the approaching squall line west of Dodge City (see the 9:16 pm CDT radar image up above).  This boundary intersection would be a favored location for a tornado from an embedded circulation at the intersection point.

So, it seems that the boundary positioning on May 6, depending on whether a particular storm had access to warm and unstable surface air or was located over the top of cold post-frontal surface air, had much to do with tornado potential.

I don't have much room here to go into Tuesday, May 7 in the Texas panhandle.  But it is worth noting that a stationary front oriented southwest to northeast appeared to have much to do with supercells' ability to produce or not produce tornadoes that Tuesday afternoon.  Storms immediately north of Amarillo at mid-afternoon appeared unable to produce tornadoes when they moved north of a fine line (not shown) marking the frontal position and colder surface air.  However, some storms located more to the northeast of Amarillo were able to access warm and unstable surface air due to their location just south and east of this same boundary:
SPC's own Roger Edwards was chasing near Amarillo that afternoon, and noted on his Twitter post to #txwx the cold air coming out of the storm he was chasing just north of the boundary, indicative of the coldness of the surface air associated with the boundary:
Here's the SPC depiction of STP at mid to late afternoon, suggesting a supportive environment for supercell tornadoes over much of the Texas panhandle, but not doing much to indicate where the surface frontal boundary and cold surface air were truly located: 
This is why it is so important to keep track as much as possible where relevant surface boundaries are located when considering which storms are more likely to produce tornadoes.

Finally, an aside on a different subject... images (below) of the EF2 tornado on the west and southwest edge of Lincoln, Nebraska on Sunday, May 5 show how difficult to classify some tornado events are.  Although radar images depicted a rotating supercell (below), images of the tornado (also below) look more like a gustnado along a storm gust front:
This high-based tornadic storm formed just ahead of a slow-moving east-west front where SPC mesoanalysis graphics indicated steep low-level lapse rates (below).  This suggests the tornado might have been some kind of a "hybrid event" where a front flank mesocyclone and the storm's gust front got together in some way to produce the tornado.

In summary, May 5, 6, and 7 had some very interesting settings for tornadoes in the Plains, and provided some excellent examples of how boundaries can sometimes help and other times hurt tornado production.

- Jon Davies  5/12/19

Friday, May 3, 2019

First large Plains tornado outbreak of 2019 on April 30 results in two deaths


Tuesday's tornado outbreak on April 30 in Oklahoma (OK), Missouri (MO), Arkansas (AR) and Texas (TX) saw many tornadoes that were widely-photographed, and unfortunately, one death from a large tornado after dark in southeast OK (see 1st image above).

*** Update 5/4/19  -  Sadly, a 2nd woman has died from the large tornado near Blue in southeast OK just after dark on April 30 -- that makes 2 tornado deaths on April 30, instead of only one. ***

For so many tornadoes (some large, some rain-wrapped), it is surprising that there were only a few injuries and just the one death (two other deaths occurred due to flash flooding in OK and MO).  That may speak to the effectiveness of warnings on April 30, and the fact that no large towns were hit directly.

The 2nd and 3rd images above show a large and dangerous EF2 tornado at mid-afternoon north of Tulsa, OK that was difficult to see at times due to rain wrapping; it is fortunate that this tornado did not strike any towns.  The 4th image above is a large EF1 tornado in northwest AR near Bergman.  And the last image above is a "landspout-type" tornado that developed in west Texas northwest of Snyder with a supercell over an east-west stationary boundary.

The outbreak was generated by a strong shortwave trough of energy at 500 mb (roughly 18,000 ft MSL) moving out of a large western trough and into the Plains (see thick red dashed line on NAM forecast for midday below).  Midlevel winds spread out widely ahead of this shortwave, inducing large-scale ascent over the area where tornadoes occurred on April 30:


The surface map at 2:00 pm CDT (1900 UTC) showed a stationary front draped from MO to west TX, with several localized areas or "waves"of low pressure:
In particular, cool air and easterly winds from overnight and morning storms over eastern KS and western MO had reinforced the front over northeast OK just ahead of the surface low there.  This increased low-level wind shear in that area, reflected by increased values of the effective-layer significant tornado parameter (STP) on the SPC mesoanalysis at 2:00 pm over northeast OK into southwest MO:

That's the area where the most tornadoes occurred during the afternoon, seen on the visible satellite image below from 3:00 pm CDT (2000 UTC), with some tornadic storms labelled:


Other storms in northwest TX near Wichita Falls, and west TX near Snyder, produced tornadoes after 4:00 pm CDT (2100 UTC) as noted on the above satellite photo.  (The last tornado image at the top of this post, northwest of Snyder TX, may have involved some landspout-type processes with strong heating along the stationary boundary and near the storm there, labelled on the satellite image.)

Moving into evening on April 30, the 8:00 pm CDT (0100 UTC) surface map and composite radar image below showed storms "lining out" across OK along the front, with some outflow boundaries hinted at over northern TX and southern OK, and also southwest MO and northwest AR:





It's possible these subtle outflow boundaries may have helped with tornado production with storms ahead of the line in the warm sector over southern OK around 9:00 pm CDT where the large EF3 tornado killed a woman between Bokchito and Blue, and also the EF2 tornado south of Springfield MO around 8:00 pm CDT where 3 people where injured.

Also notice how the effective-layer STP from the SPC mesoanalysis highlighted southern OK and southwest MO at 8:00 pm CDT, largely due to an increasing low-level jet (southerly winds at about 5000 ft MSL, not shown) within the warm sector around sunset:

This outbreak had a little of everything regarding tornadoes.  It involved tornadoes occurring well inside the open warm sector (the infamous drone-viewed tornado near Sulphur OK around noon), tornadoes near convectively-enhanced and shear-enhanced boundaries (the afternoon tornadoes in northeast OK, southwest MO, and northwest AR), the deadly evening tornado in southeast OK within the warm sector (possibly helped by a subtle outflow boundary), and the aforementioned afternoon tornado in west TX that may have involved landspout processes. 

- Jon Davies  5/3/19