Thursday, May 13, 2021

The May 8, 2021 HP supercell in central Kansas & comparison to the May 1, 2018 Tescott tornado setting

Last Saturday's HP supercell at early evening west of Salina (see Shawna's dynamic photo above) was our first chase in nearly a year, as Kansas (KS) has been rather quiet concerning severe weather recently.  The setting did have some similarities to the May 1, 2018 tornadic supercell set up northwest of Salina three years ago in that storms formed in west-central KS at the north edge of a "cap" (warm layer of air aloft near 10000 ft MSL), and in each case at least one storm moving east-northeast became an evening supercell near Ellsworth, KS.  

But Saturday's supercell was non-tornadic and high-precipitation ("HP") in character, while the May 1, 2018 supercell produced an EF3 tornado on the ground for around 14 miles.  Sometimes it is informative to compare cases in the same geographic area that appear to have similar settings at first glance, but end up with very different results, which I'll do briefly here, as well as summarize our storm chase.  

The NWS-analyzed surface map at 7:00 pm CDT (0000 UTC 5/9/21) in the first panel below showed a double-barrel surface low near a triple point over west-central KS, northeast of which storms had formed at mid to late afternoon... a classic severe weather setting  sometimes associated with tornadoes:

The 2nd panel above also shows the NAM model 12-hr forecast of the 0-1 km energy helicity index (EHI, combining instability and low-level wind shear) valid at the same time as the surface map in the first panel.  EHI values approaching 2.0 into central KS suggested that tornadoes might be a possibility, along with the morning SPC outlooks (not shown) that showed a 5% probability of tornadoes in central and north-central KS... not high but not insignificant, either.

Shawna and I followed a couple cells that had formed west and southwest of La Crosse, and had became a loose cluster of storms as they moved north of Great Bend.  Out of this "cluster", a high-based supercell formed after 7:00 pm CDT west of Ellsworth, KS:

This storm moved across Ellsworth producing wind gusts, heavy rain, and some nickel-size hail, and we were able to get in front of it again on Hwy 140 heading east toward Brookville and Salina.  Here's Shawna's shot of this high-based cell near Carneiro looking west-southwest as I snapped some pics in the foreground:

This HP storm really took on some impressive structure west of Brookville, with a dark roiling high-based shelf (the storm was outflow-dominant at this point, with winds already shifting to the west at our location in front of the shelf).  Staccato cloud-to-cloud and cloud-to-ground lightning bolts were visible along the outflow shelf, and I took the picture below of this "angry" supercell looking north (similar to Shawna's photo at the top of this post, but without the cool lightning) from west of Brookville as the storm approached us:

The high outflow base and outflow-ish nature of the storm eliminated thoughts of tornadoes at this point, but it was a beautiful storm in a pretty setting (the Smoky Hills).  We reached Salina with darkness upon us, and after watching lightning for a while, we headed back to Kansas City.

One problem working against tornado potential on this day appeared to be high lifting condensation level (LCL) heights (up near 2000 m above ground) that allowed evaporative cooling as rain fell through the deep sub-cloud layer, generating cool outflow air that undercut the storm's rain-wrapped mesocyclone.  Another issue was somewhat weak winds aloft in mid-levels that generated just enough deep-layer shear (25-30 kt) to support a supercell, but one that was HP in character, adding to the outflow-ish nature of the storm.  The morning HRRR model forecast these limitations rather well:

But the morning HRRR model also did a good job forecasting the supercell at early evening, with a mid-level rotation track indicated west of Salina:

As mentioned earlier, the track and timing of this supercell were similar to the May 1 Tescott tornadic storm back in 2018.  However, there were some key differences in the environments, as indicated by the red highlight boxes in the table of parameter comparisons from SPC mesoanalysis graphics (not shown) between the two events listed below:

Apart from smaller low-level wind shear (0-1 km storm-relative helicity or SRH) on May 8, the main differences were much higher LCL heights compared to May 1, 2018, and smaller deep-layer shear (0-6 km bulk wind difference or 6BWD), confirming the same issues raised by the HRRR model forecast discussed earlier.

Here's a RAP model analysis sounding (similar to what is used in generating SPC mesoanalysis graphics) at Ellsworth, KS valid at 7:00 pm CDT on May 8, 2021, highlighting in yellow and red these same problematic issues in the environment regarding tornado potential at the time the supercell was ramping up west of Ellsworth:

This sounding confirms that LCL heights and deep-layer shear were two negative factors in the environment that probably worked against tornadoes on May 8.

As the evening went on, low-level shear increased (SRH, not shown) and LCL heights lowered quite a lot (not shown) as surface temperatures cooled.  But, low-level stability (convective inhibition or CIN) also increased (not shown), suggesting that storms were not strongly surface-based near and after dark, so that much of the instability was coming from well above the ground, working against the formation of tornadoes.

Before closing, it is interesting to look at a later RAP model analysis sounding for Salina at 10:00 pm CDT (0300 UTC 5/9/21).  This sounding was estimating the environment in front of a 2nd supercell that formed behind the one we chased earlier, and was tornado-warned based on radar-detected mid-level rotation around 10:11 pm CDT:

Notice how large the CIN (blue area on the thermodynamic diagram) had become after dark, with more than 225 J/kg of inhibition.  This was the result of outflow air behind the earlier supercell, now bowing out well east of Salina (SLN) on the radar image below, shortly before 10:00 pm CDT:

I've indicated on the radar image the new tornado-warned cell west of Salina and the "fine line" showing the trailing outflow boundary from the earlier supercell (white arrows).  Also note the temperature and dew point at Salina (63 over 59 deg F) within this outflow just before 10:00 pm CDT.  All this information suggests that this new supercell was ingesting cool outflow air that would undercut and eliminate any true tornado potential with this new supercell.

I was a little surprised that NWS issued the tornado warning, based on radar and environmental data above.  Although ping-pong size hail fell with this 2nd supercell, no tornado occurred and the warning was soon dropped.  This re-affirms how outflow near the ground can really interfere with tornado production.  

The May 8, 2021 central KS case is an interesting case to study regarding how factors like LCL height and relatively weak deep-layer shear can limit supercell tornado potential.  These are two easy-to-overlook ingredients that require careful consideration in tornado forecasting.

- Jon Davies  5/13/21

Thursday, April 1, 2021

March 25, 2021 deadly tornado outbreak in Alabama and Georgia

One week ago today, a deadly tornado outbreak in Alabama (AL) and Georgia (GA) caused 6 deaths and a number of injuries, making it the deadliest tornado episode of 2021 so far.

Similar to March 17, 2021 (see toward the end of my post here), the outbreak was well-forecast and well-warned, and as a result many lives were likely saved, even considering the deaths that did occur.  As a result, some people on Twitter have mentioned the famous April 27, 2011 tornado "super outbreak" ten years ago, when over 200 tornadoes occurred on one day in the Dixie states and more than 300 people were killed.  

The graphic at the top of this page offers a quick comparison of the tornado environments between that day back in 2011 and the recent outbreaks in AL on March 25 and March 17, 2021 using a scatter-diagram showing storm-relative helicity (SRH, a measure of low-level wind shear) and mixed-layer CAPE (a measure of instability).  Sizable low-level wind shear can generate low-level storm rotation leading to supercell tornadoes, and CAPE (convective available potential energy) helps support and sustain those rotating storm updrafts.

The big pink dot in the middle of the diagram up at the top of this page is representative of the environment on April 27, 2011, showing an unusually large combination of SRH and CAPE leading to the massively deadly tornado outbreak that day nearly ten years ago.   It's pretty clear that 4/27/11 had much larger support for strong and violent tornadoes than either 3/25/21 or 3/17/21.

Getting back to 2021, I've plotted on the same diagram four of the strongest tornadoes from March 25 last week (small yellow-filled dots) and four of the strongest tornadoes from March 17 a couple weeks ago (small red-filled dots), and enclosed those areas on the diagram in black ovals.   Comparing 3/25/21 with 3/17/21, it appears that March 25 had somewhat larger low-level wind shear, one difference between the 3/17/21 setting that supported shorter-tracked EF2 tornadoes, and the 3/25/21 setting where tornadoes were stronger (EF3 and EF4) and longer-tracked.

Another difference between March 25 and March 17 (not shown here) was deep-layer wind shear, where 0-6 km bulk wind difference (BWD) on 3/25/21 in central AL averaged near 70 knots or more, whereas on 3/17/21 the 0-6 km BWD averaged only near 50 knots,  In other words, the general wind fields and resulting wind shear at multiple levels appeared stronger on March 25, 2021, helping to generate stronger and longer-tracked tornadoes like this one near Greensboro AL and west of Centreville AL at mid-afternoon:

To very briefly review the setting on March 25, here's the tracks of EF2+ tornadoes last Thursday plotted on top of a surface map at 2100 UTC (4:00 pm CDT) on 3/25/21:

And here's an SPC mesoanalysis graphic of composite tornado parameters at 1800 UTC (1:00 pm CDT):

The warm sector over central Alabama south of the warm front was where wind shear and instability parameters were maximized and the long-track afternoon tornadoes occurred.

Here's a composite radar image at 1930 UTC (2:30 pm CDT) just before the town of Ohatchee in east-central/northeast AL was hit by a tornado that killed 5 people.  

And here's a very interesting photo as that killer tornado approached Ohatchee around 1940 UTC (2:40 pm CDT) -- it appears to show two large vortices, reminiscent of the multi-vortex Elkhart Indiana tornado photographed on Palm Sunday in April of 1965:

This tornado was from the same supercell that produced a 50-mile track EF3 tornado that struck earlier south of Birmingham with damage and injuries.  

And, on the radar image earlier, a new supercell is indicated entering west-central AL that would later generate a strong 80-mile track EF3 tornado starting west of Centreville, AL that at times was more than a mile wide (see the tornado photo earlier by Max Olson):

Here's a RAP model mid-afternoon analysis sounding at Centreville, AL showing the environment supporting this longest-track tornado of the day that could have possibly been rated EF4 had it directly hit a town:

With around 300 m2/s2 of 0-1 km SRH, 1400 J/kg of mixed-layer CAPE, over 80 kt of deep-layer shear, and sizable low-level CAPE, this was a fairly potent environment for supporting supercell tornadoes.

As the outbreak extended into evening on March 25, a new storm became tornadic in central AL around 9:00 pm CDT (0200 UTC 3/26/21, not shown).  On the graphic below, the wind and instability environment for this storm at 0300 UTC 3/26/21 on SPC mesoanalysis composite tornado parameters was still quite good for supporting strong or violent supercell tornadoes:

And that's in fact what happened, with a low-end EF4 tornado striking Newnan GA over the border from AL around 0400 UTC (11:00 pm EDT), killing one person.  After that, storms tended to outrun the warm sector instability axis as they moved eastward, and tornadoes ceased occurring.

Finally, to back up in scale and time a bit, here's the NAM model 9-hr forecast of 500 mb features and energy-helicity index (EHI, combining SRH and CAPE) at mid-afternoon on March 25:

As on March 17, it was again a classic setting with plenty of SRH and CAPE in an area ahead of a strong trough at 500 mb (this time, a strong mid-level shortwave) where lift and forcing resulted from a "spreading" jet branches pattern ahead of this mid-level disturbance.

Last Saturday March 29, 2021 saw yet another deadly tornado day within an active sequence of days, with tornadoes in Texas, Louisiana, Arkansas, Tennessee, and Mississippi; one person was killed in northeast Texas (see photo of EF2 tornado near Carthage, Texas below):

As I write this, it looks like conditions supportive of tornadoes will take a break and hold off until the middle of next week (around April 7th or 8th?) in the central and southern U.S.

-  Jon Davies 4/1/21  (no April Fooling)

Wednesday, March 24, 2021

Active March severe weather pattern continues across the central and southern U.S.!

After last week's St. Patrick's Day outbreak in the South on March 17 (more about that farther down), yesterday saw yet another cold-core type setting with tornadoes in west-central Illinois (photo above) and southeast Iowa, on top of last week's cold-core activity (see my last blog post here).

Tomorrow (March 25, 2021) will likely see another tornado outbreak in the southern Mississippi River Valley into parts of Mississippi, Alabama, Tennessee, and Kentucky.  This will probably be similar in scope to the March 17 outbreak with several large long-track tornadoes, so people living in those areas should monitor weather information carefully!

Yesterday's cold-core event was weak (2 EF0 tornadoes), but fairly classic pattern-wise, occurring in the vicinity of a Pacific cold front intersecting a warm front in the northeast Missouri (MO), west-central Illinois (IL), and southeast Iowa (IA) area east of a closed mid-level low.  That was unlike last week's cold-core activity in Kansas, Nebraska, and Missouri (see blog post here) when the tornadoes occurred along occluded sections of a frontal boundary near the mid-level low, instead of the more common Pacific front/warm front intersection area. 

Here's the 0000 UTC 3/24/21 surface map (7:00 pm CDT 3/23/21) shortly before the tornadoes yesterday:

The tornadoes occurred near Rockport, IL (photo at top of this page), and Ft. Madison, IA, marked on the surface map above, close to the Pacific cold fornt / warm front intersection as mentioned earlier, roughly matching the Fig. 16 composite in Davies (2006).

The visible satellite photo at the same time (tornadic cell locations at arrows) showed the cloud swirl associated with the mid-level closed low at 700 mb (roughly 10,000 ft MSL) and 500 mb (roughly 18,000 ft MSL):

The 12-hr 700 mb forecast from the NAM model morning run on 3/23/21 showed the mid-level low over southwest IA, and a broad area of cold air aloft well below freezing (the blue colors) spreading over the warm sector into western IL:  

With surface dew points in the mid-50's (deg F) near the warm front, that cold air aloft generated plenty of instability to support storms and some brief tornadoes.

The 9-hr HRRR model forecasts from mid-morning suggested the location of the Pacific front/warm front intersection fairly well from surface temperatures and winds (1st panel below).  Plentiful low-level CAPE (0-3 km, 2nd panel) was also forecast to surround the boundary intersection area and extend northward into southeast IA:

The low-level CAPE north of the warm front into IA may have been a contributor to the brief tornado with minor damage near Ft. Madison, IA that occurred some distance north of the warm front within a linear feature on radar (not shown), unlike the tornadic supercell near Rockport, IL.

Back on March 10 (2021), there was a cold-core setting that affected south-central Minnesota that was _not_ tornadic, although a photogenic supercell (not shown) occurred near the warm front in that case.  Several people have asked me why that setting did not produce tornadoes, and while I don't have definitive answers, comparing that evening's surface map with yesterday's surface map suggests some clues.

Notice on the 3/10/21 surface map at 2300 UTC (5:00 pm CST, shown below; location of tornado-warned supercell marked by circled "S") that the warm front was "bulging" much farther northward relative to the surface low:

As a result, comparing this map with the surface map earlier from yesterday, there was not much of a "spiraling" wind flow pattern into the surface low.  Instead, winds went from a south or south-southeast direction across southeast Minnesota and most of Wisconsin abruptly to a north or north-northeast direction in the area northeast of the surface low.  This was different than yesterday's surface setting (see the surface map shown earlier) where winds from the warm sector over Illinois and then flowing into the Iowa surface low (ignoring a few "outflow" locations) backed more gradually across the warm front and farther northwestward, going from southerly to southeasterly to easterly.  This more "spiraled" backing wind flow pattern into the surface low may have helped set up a better low-level shear environment for low-level rotation in yesterday's storms near and just north of the warm front.

I'll make one other observation comparing the two surface maps.  In the non-tornadic March 10 case, the temperature contrast across the Minnesota warm front was > 10 deg F in a short distance (say 10-15 miles), which would likely have a detrimental effect on low-levels for tornado potential with the supercell that was crossing the warm front that day.  Yesterday, the temperature contrast across the warm front and for miles north of it was rather gradual, only around 5 deg F between west-central IL and southeast IA, and would not have as abrupt an impact on storms crossing the warm frontal area.

Before I close, I wanted to talk briefly about last week's St. Patrick's Day tornado outbreak in the South.  It was well-forecast days in advance by SPC, announced aggressively in media, and there were  no deaths and only a few injuries in spite of 30+ tornadoes occurring!  The excellent forecasts and media coverage appeared to really contribute to the low injury toll and the lack of deaths.  I truly hope the same will be true with tomorrow's expected outbreak in some of the same area.

Briefly, regarding the 3/17/21 outbreak, here's the morning NAM model forecast at 500 mb valid at mid-afternoon, along with an inset of the NAM 0-1 km energy-helicity index (CAPE and low-level shear combined) for the same time:

The setting was fairly classic, with a strong trough moving eastward across the southern U.S., and "spreading" jet branches providing forcing for generating storms out ahead of the trough where CAPE/shear combinations were strong and very supportive of tornadoes across the broad warm sector in the Dixie states.   

Notice too, that there was yet another "cold-core" setting associated with the same system, this time over southwest MO, where some weak tornadoes occurred.

And getting back to the Dixie states, below are a couple images of a supercell and long-track tornado (note the horizontal vortex on the second image) that moved across southwest Alabama (see tornado track on inset in image above) at mid-afternoon.

It is a little unusual to be able to see such visible supercell structure in the South.

I hope people in the Dixie states and the southern Mississippi River Valley/Ohio River Valley areas pay careful attention to weather information tomorrow, March 25, 2021.  Stay safe!

- Jon Davies  3/24/21

Tuesday, March 16, 2021

March 13, 14, and 15 bring big tornadoes and cold-core action to the Plains


Friday, March 12 through Monday, March 15, 2021 saw tornado activity really ramp up in the Central Plains.  There were no deaths or injuries reported, as the biggest day (Saturday, March 13) was well-forecast and warned.

Over 10 tornadoes occured on March 13 in the central and southern Texas (TX) panhandle, including the twin large tornadoes pictured above from the same supercell near the town of Happy, along with a storm-relative velocity image at roughly the same time showing the two mesocyclones side by side.  The tornado and mesocyclone labeled "1" above generated a tornado rated EF2 on the ground for 17 miles according to NWS Amarillo (had the tornado hit a town, the EF rating would likely be higher).

Farther east, according to radar, it appeared there were also two tornadoes on the ground simultaneously at times with the Happy, TX supercell according to radar (not shown).  But these were not photographed as the storm became increasingly wrapped in rain and the radar circulations occurred in difficult spotter/chaser terrain northeast of the Palo Duro Canyon State Park.  It doesn't appear that there were any tornado mergers, but that's not entirely clear, and there were also times when multiple mesocyclones appeared to be interacting with each other (not shown).

The following two days, other localized tornadoes occurred in northwest Kansas (KS) and southwest Nebraska (NE), including the "white" landspout tornado pictured below (1st photo) west of Trenton, NE on Sunday March 14, and a tornado near Stilwell, KS, south of Kansas City (2nd photo) where some minor damage occurred (rated EF0):

I'll briefly touch on the meteorological setting for March 13, March 14, and March 15, as these days involved a closed 500 mb low within a large mid-level trough moving slowly northeastward from the southwestern U.S.  On March 13, the 500 mb low was too far to the west to be directly involved in the Texas tornado outbreak, but on the subsequent two days, the cold-core low aloft had a direct impact on tornadoes occurring in rather marginal settings.

Below is a visible satellite image at mid-afternoon on March 13, with surface weather map features drawn in, and the location of the 500 mb low well to the west:

The tornadic supercells occurred in the warm sector ahead of a Pacific cold front and south of a warm front over the TX panhandle, including the Happy, TX supercell (labeled with arrow).   Below are SPC mesoanalysis images of two composite tornado parameters, suggesting how favorable for supercell tornadoes the environment was over the panhandle at 2200 UTC (4:00 pm CST) while tornadoes were ongoing near and east of Happy, TX:

And here's an analysis sounding from the RAP model at Plainview TX (south of the Happy, TX supercell) at 2100 (3:00 pm CST) at about the time the outbreak started:

Notice that parameters such as mlCAPE (> 2000 J/kg), 0-1 km storm-relative helicity (SRH > 250 m2/s2), surface to 6 km shear (> 70 kt), and 0-3 km mlCAPE (> 150 J/kg) were all highly favorable in this environment for supercells capable of generating strong tornadoes.

The next day (Sunday, March 14), satellite with surface features (below) showed that the 500 mb low had moved to eastern Colorado (CO), where a blizzard dumping over 2 feet of snow was in progress over northern CO and southeast Wyoming.  The surface low had also evolved northward to eastern CO, with a massive dry slot/clear slot punching northward across most of Kansas with the 500 mb low, while the Pacific cold front had advanced eastward:

Along the occluded front extending northeastward from the CO surface low, the wind shift along the boundary combined with surface heating and CAPE (from the cold air aloft with the mid-level low) to help produce a couple of brief landspout tornadoes at early afternoon, including the Nebraska EF0 tornado pictured earlier.

An SPC image of 0-3 km mlCAPE and surface vorticity at 1900 UTC (2:00 pm CDT) showed very well the low-level CAPE (in red) overlapping the west-east boundary (blue surface vorticity lines) over northwest KS and southwest NE:

These are ingredients that can generate non-supercell tornadoes such as landspouts when low-topped storm updrafts from CAPE and heating phase properly with the boundary so that vertical vorticity along the wind shift can be stretched upward.

Here's a RAP model 1-hr forecast sounding near Trenton, NE  shortly before the brief NE tornado pictured earlier -- notice how the CAPE is relatively small but bunched low in the  atmosphere, typical of tornado settings near 500 mb lows:

And here's an image showing the crisp low-topped storms looking north from KS into NE associated with this occluded front and cold-core setting:

Last, here's the satellite/surface setting on Monday, March 15 between 2000 UTC and 2100 UTC (3:00-4:00 pm CDT) over the KS-Missouri(MO) area:

The 500 mb low had "split" or evolved into two centers, with one drifting east across Kansas and generating a surface low over northeast KS near the stalled occluded front.  With all the cold air aloft associated with the 500 mb low and the west-east boundary near Kansas City, the stage was set for a couple more brief tornadoes, including the one pictured earlier near Stilwell, KS.

The SPC depiction of low-level CAPE and surface vorticity at 2100 UTC showed a setting similar to the day before (low-level CAPE overlapping the boundary ), except this time near Kansas City:

And here's a RAP model 1-hr forecast sounding at Harrisonville, Missouri, shortly before the tornado at Stilwell, KS just west across the KS-MO border:

With a somewhat more clockwise-curving hodograph, the Stilwell storm appeared more supercellular on radar (not shown), but the components of low-level CAPE and the boundary were also likely significant contributors to the tornado in this occluded front and cold-core setting.

In the March 13, 2021 Texas case, parameters and ingredients were strong within the warm sector away from pre-existing fronts and boundaries to help generate several classic-type supercell tornadoes.  But on March 14 and 15 as the 500 mb low moved slowly out into the Plains, the tornadoes were more a result of low-topped storms interacting with an occluded front, where the boundary was a key factor.

It is interesting to note that the Texas twin tornadoes pictured at the top of this article occurred 31 years to the day after the twin Hesston-Goessel KS tornadoes that went on to merge together into an EF5 tornado.

And I must note that St Patrick's Day 2021 (tomorrow) will likely see a significant outbreak of severe weather and tornadoes in the South over the Arkansas-Louisiana-Mississippi-Alabama area.  I hope everyone in those areas stays informed and up-to-date on the weather.. stay safe!

- Jon Davies  3/16/21

Monday, February 22, 2021

A difficult tornado to forecast - 2021's 2nd deadly tornado on February 15 kills 3 people in North Carolina

Last Monday's EF3 tornado near the coast of southeast North Carolina (north of Sunset Beach NC) killed 3 and injured 10.  Because it occurred at night (between 11:30 pm and midnight EST), there were no photos of the tornado, but a storm chaser managed an image of the tornadic supercell (top photo above) off the coast east of Myrtle Beach, South Carolina (SC) around 30 minutes before it struck on land just north of the NC-SC border.   The 2nd photo above shows some of the impressive damage (see this page on the tornado from NWS Wilmington NC).

There was no tornado watch in NC ahead of the storm, and a tornado warning was not issued for Brunswick County where the tornado occurred until 5 minutes after it developed.  That's because it was far from an obvious forecast setting, which is worth posting about here.  The environment was also evolving rapidly, and the tornado developed fast on radar (not shown) when the supercell moved on land.

A forecast and nowcast problem in this case was that surface-based (sb) instability at 11:00 pm EST 2/16/21 (0400 UTC) appeared to be well off shore from the Carolinas (see sbCAPE on the first panel below from the SPC mesoanalysis):

When there is no surface-based instability within a severe weather threat area, storms that do occur are considered to be "elevated", meaning that the instability (CAPE) supporting the storms comes from lifted air located somewhere above the ground.  Because tornadoes by definition occur at the ground (different from a "funnel" aloft), tornadoes are generally not expected in such elevated settings because air at the ground is typically too stable to support tornadoes.

This "elevated" factor affected composite tornado forecast parameters such as the effective-layer significant tornado parameter (STP), shown in the 3rd panel of the SPC graphic above, where effective STP was near 0 along the coast of the Carolinas, a result of this apparently "elevated" environment. 

However, looking at mixed-layer CAPE (mlCAPE, 2nd panel in the graphic above), which uses an average of the temperature and moisture properties within the lowest 1 km above ground, notice that some CAPE (around 500 J/kg) was present right along the northeast SC and southeast NC coast at 0400 UTC near Sunset Beach (black square dot).  This is because there was significant moisture and increased temperature flowing northward above a shallow stable layer at the ground.  A RAP model sounding profile at Southport, NC (roughly 25 miles east of Sunset Beach), modified using a 0430 UTC temperature and dew point surface observation of 64 deg F and 64 deg F, showed this:

Notice  that the yellow dashed lifting curve at left on the graphic above (representing a surface-based lifted parcel of air) generated _no_ CAPE, suggesting an "elevated" environment.  But in contrast, the yellow dashed lifting curve at right (representing a lifted parcel of air from around 0.3 km above ground), did indicate notable CAPE, as seen by the red mlCAPE area on the sounding.  

In other words, significantly unstable air was present less than 1000 ft above the ground, just above the very shallow stable layer.  It was this instability, combined with large low-level wind shear (storm-relative helicity or SRH of 350-400 m2/s2) near a warm front that helped support low-level rotation and a tornado in the supercell that moved on shore near Sunset Beach NC.  Apart from the shallow stable layer, the environment was a small instability and large wind shear setting often typical of southeast U.S. tornadoes in the cool season.

Warm fronts with significant combinations of CAPE and SRH along and near them are recognized in forecasting as potentially supportive of tornadoes with supercells that move along with or across them.  The surface map below at 11:00 pm EST (0400 UTC) shows that, indeed, a warm front was present moving northward near the NC-SC border along the coast:

Also, the SPC graphic below at 0400 UTC showed that some low-level instability (mlCAPE in the lowest 3 km above ground) and large low-level wind shear (SRH in the lowest 1 km above ground) were both present in a small corridor along the NC-SC coast, near the warm front seen on the surface map above:

Although these ingredients were present only in a narrow strip along the coast, they apparently were enough to support the deadly supercell tornado near the warm front and the coast.  But the shallow stable layer near the ground discussed earlier masked this potential by suggesting an elevated environment along the same warm front that was actually not as significantly "elevated" as it looked at first glance.  This was due to the shallowness of the stable layer along the coast and the warmth/moisture flowing northward just above the ground.

Radar below shows the supercell just off shore at 10:00 pm EST (0300 UTC) that eventually moved on shore in NC near Sunset Beach around 11:30 pm EST (0430 UTC), interacting with the surface warm front and CAPE/wind shear just above the ground to produce the deadly tornado that lasted until midnight EST:

As the supercell moved farther inland away from the coast, the low-level stable layer apparently became too deep, eliminating near-surface CAPE, and there were no other tornadoes over eastern NC.

Forecast maps from the morning of February 15 did hint that there might be ingredients for a marginal possibility of tornadoes near the SC-NC coast, but that potential was certainly not very evident.  Below are the NAM model 500 mb forecast for mid to late evening, and a 2-panel NAM model forecast of mlCAPE and 0-1 km SRH:

At 500 mb (roughly 18,000 ft above sea level), a large trough was moving through the mid-section of the U.S., bringing with it last week's news-making frigid arctic cold to the central Plains and particularly Texas.  Ahead of this trough, a spreading jet branch pattern was over the southeastern US providing lift through diffluence, and at mid-evening some mlCAPE was forecast just touching the NC-SC coast, barely overlapping large 0-1 km SRH near the afforementioned warm front.  This narrow  juxtaposition of ingredients was a hint of sorts, but only a small one and nothing that appeared very threatening or alarming for the evening forecast environment. 

Earlier on 2/15/21 (not shown), an EF2 tornado struck southwest Georgia at mid-afternoon, also located near the same warm front that traveled rapidly during the day from the Florida Panhandle all the way to the Carolinas. The warm front in essence outran the instability axis as it moved across eastern Georgia and into South Carolina during the late afternoon and early evening, but then encountered a new fetch of instability again by late evening coming off the Atlantic Ocean along the coast of the Carolinas when the deadly tornado occurred.

Warm fronts are a feature to watch carefully in tornado forecasting, particularly fast-moving ones as in this case.  Often the environment in the immediate vicinity of the warm front may appear "elevated" with a layer of cool surface air north of the front.  But if the front is moving rapidly, the environment may be changing rapidly as well with warm/moist near-surface air flowing quickly northward, producing CAPE where there was none an hour or two earlier.  

This case also emphasizes that when a cool near-surface layer is very shallow along a warm front, sufficient instability may be present to support tornadoes even if the setting appears "elevated" at first glance.  Rapidly-moving warm fronts can really modify and change a local environment quickly.

- Jon Davies  2/22/21