Wednesday, April 10, 2013

04/08/13 damaging nighttime tornado in swrn NE with strong CIN

On 04/08/13, a cluster of small updrafts were ongoing N of Goodland KS around 01Z.  Between 0140Z and 0200Z, two updrafts exploded on the srn flank of this activity, turned to the right of the mean wind, and per LSRs issued in the next 18 hours apparently produced simultaneous damaging supercell tornadoes 3E Benkleman NE and 8SSE Max NE near 03Z.  Of primary importance, the storms were likely in an environment characterized by rather strong inhibition to surface-based parcels. **Edit: recent PNS from WFO GLD officially verified that the westernmost of the two supercells did produce a tornado.  It occurred at 0310Z, lasted one minute, and was rated EF-1.  The natural conclusion would be that the tractors flipped 8SSE of Max by the easternmost supercell were deemed to be caused only by non-tornadic winds.**

Monday afternoon, an energetic upper trough--with most of its jet energy on the back side--was digging through the wrn CONUS with 543-dm 5H closed low centered over swrn UT.  Associated sfc pressures were anomalously low from the 4 corners states swwd thru the lwr CO river valley; a 988 mb sfc low was centered near PUB, and sfc pressures as low as 991 mb reached the srn half of AZ (which is in excess of 5 standard deviations below climatological  norms for that area; i.e., very, very rare for early April).  Forcing for upward vertical motion likely ejected across the central High Plains by late afternoon in support of convective development.  A diffuse dryline would be the focus for initiation, with the segment from some distance NW of AKO (where in intersected a sagging cold front; see rough sfc analyses below) to 50SW GLD looking to have the highest probability of initiation based at least partly on background modulation on the synoptic scale.

A rather veered nocturnal LLJ acted to shallow the return-flow moist layer somewhat over the srn Plains by Monday morning, with boundary layers only 1 km deep at DFW/OUN capped by deep/seasonably warm EML layer immediately above.  Throughout the day a modest sely LLJ re-developed over KS/srn NE, as the primary LLJ TX-OK-MO gradually backed... contributing to moisture advection E of dvlpg/effective dryline.  By 00Z a 1250 m moist layer was observed at DDC (although this was co-located with the dryline); while 1500 meter depths were observed at LBF (just south of aforementioned cdfnt) and LMN; and a nice 1900 m moist layer depth was observed at TOP.

The 00Z LBF RAOB's T/Td profile looked gorgeous, with backed mid-level trajectories during the day having maintained cool capping aloft (e.g. 1 C at 7H) above the boundary layer, and 68/55 F at sfc contributing to ~2000 J/kg MLCAPE and ~0 J/kg MLCIN; and moreover, a profile that would support a weak CIN environment even with diurnal cooling of the near-surface layer.  However, the DDC RAOB showed 7 C at 7H, as did the DEN RAOB with its deeply mixed sfc-based layer.  Model initializations/mesoanalyses also indicated the tightest gradient in the capping aloft (e.g 3-6 C at 7H) was hugging the NE-KS border region late aftn-evening, with trajectories there resulting in WAA having developed in the capping layer.  In fact, mesoanalysis time series showed 5-6 C at 7H consistently in the near-storm environment of the small storm cluster that moved from 40ESE LIC to 20N MCK and ultimately produced the damaging nighttime tornado.

In Davies and Fischer 2009 (, hereafter DF09, 1705 RUC soundings associated with non-tornadic, weak tornadic, and significantly tornadic supercells were examined from a thermodynamic and vertical shear standpoint to see which parameters might best discriminate among the categories both during the day and at night.  Ultimately, MLCIN proved to be the strongest discriminator among nighttime non-tornadic supercells and nighttime significant tornadic supercells... with almost no overlap in the middle 50% of each category, a strong overall difference from the MLCIN distribution within the daytime subset.  MLCIN of less than 50-75 J/kg was very common among nighttime weak and nighttime significant tornadoes.  In fact, 90% of the nighttime significant tornadoes occurred with less than 75 J/kg MLCIN.  (These overall findings are not surprising when viewed in the context of Davies' groundbreaking paper on CIN & LFC heights associated with tornadic and non-tornadic supercells:  0-1 km SRH also showed a pretty dramatic increase when moving from nighttime nontornadic supercells to nighttime significant tornadoes, especially compared to the daytime subset, but a lot of overlap occurred among the nighttime categories.

In a companion paper (Fischer and Davies 2009 at, hereafter FD09, we looked carefully at two long-lived supercells in rather strong CIN environments which produced 2 and 4 nighttime sig tors respectively: the 5/29/08 Jewell and Belleville KS area tornadoes; and the 06/11/08 Salina, Champan, Manhattan, and Soldier KS area tornadoes.  Additionally... the 04/24/08 Beloit KS sig tor and 4/09/11 Storm Lake IA area sig tors had a lot of similar storm-scale and environmental characteristics to those storms in FD09.  All of these events involved well-established long-lived supercells which started out generally in rather strongly/deeply mixed environments, and toward darkfall progressed fully into the moist axis as the boundary layer diurnally cooled.  All 4 events' near storm environments possessed 1500-3000 J/kg MLCAPE, 80-140 J/kg MLCIN, 500-800 m2/s2 0-1 km SRH, 6.5-11.5 0-1 km EHI, and 60-80 kts 0-6 km bulk shear.  In all cases the CIN resulted primarily from diurnal cooling beneath a warm EML base (i.e., moisture was seasonably favorable and reasonably deep at sig tornado time).  Very largely-looped and long hodographs developing in the near-storm environment of the well-established supercells likely contributed to intense vertical pressure gradients such that the stable layer was able to be overcome to allow for production of significant tornadoes in spite of larger MLCIN than is typical of the majority of sig tor near storm environments.  While the 0-1 km SRH and 0-1 km EHI distributions in DF09 were each by far the largest in the nighttime sig tor category compared to the other 5 categories, the 0-1 km SRH and 0-1 km EHI within the two FD09 events/other two aforementioned events easily exceeded the 90th percentile of each for nighttime sig tors in DF09; i.e., off the charts, so to speak, and climatologically rather rare.

Given the type of set-up on Monday, a fairly quick transition from high-ish LCL/weak CIN environment to low LCL-strong CIN environment was expected--not strongly different from the environment evolutions in FD09 as described above.  Though backing a moist boundary layer up onto the High Plains is typically a big--and underappreciated--positive for retaining weak-CIN after dark environments in the Great Plains (given that elevation minimizes the impact of the EML)... the lack of richer/deeper moisture, and the lack of cool capping aloft in and upstream of the area, far overwhelmed the elevation impacts.  Sfc temps fell pretty quickly in the "moist sector" too, despite the strong low-level gradient...with 65-70 F by 23Z invof IML-MCK-CBK-GLD.

The 02Z RAP initialized sounding (see below) was favored over that at 03Z as the near-sfc T/Td/wind details were far more accurate on the 02Z sndg.  (Given I was away from work, I used Earl Barker internet data and thus soundings & hodographs are not modified).  With a 1 km deep moist layer with mean (ML) dew point of 56F, topped by a sharp (i.e., dry adiabatic nearly all the way down) EML base warm enough such that 7H temps were around 6 C... MLCIN of 145 J/kg was accumulated via the sounding.  The similarities of this sounding's shape with the two events' proximity soundings from FD09 are underlined by the fact that all three had a ML LFC between 2900 and 3000 meters.  The striking differences between this case and the FD09 cases, however, aside from the simple fact that this case's damaging tornado received only an EF-1 rating, are: 1) modest instability, with under 1000 J/kg MLCAPE; 2) a rather young storm, being less than 90 minutes old at the time of the tornado; and 3) somewhat weaker low-level shear, with 0-1 km SRH near 400 m2/s2 (the impact of a dramatically weaker observed motion vs. Bunkers overwhelmed the impact of a more rightward observed motion vs. Bunkers--result in this case being weaker-than-shown SRH, for what it's worth). The fact that the tornado was brief and the storm was not cyclic would tend to make sense.  The storm was also about to cross the sagging cold front, though with 1) slow storm motions and 2) (primarily non-NWS) sfc obs immediately N of the sfc front indicating (away from rain-cooled areas) quite similar sfc theta-e as just S... a further increase in CIN due to crossing the front may not have occurred particularly quickly.

-poorly analyzed 00Z and 03Z cntrl High Plains sfc analyses (obs via Al P.); red dots = rough positions of 01-02Z tornadoes on 00Z map... and of ~03Z EF-1 tornado E of Benkelman, damaging wind/false tornado report SSE of Max, and supposed "snownado" W of Akron on 03Z map
-low res radar evolution 02-03Z; recall westernmost supercell (which passed just nw outside of the tor warning polygon), whose appearance at least via base reflectivity was the less impressive of the two, was the one that produced the tornado
-02Z RAP 00-hr fcst sounding at McCook NE.  Note the lifted parcel path shown on the sndg is via a SB parcel, though a lowest 100 mb ML parcel would be more appropriate.  (Given the shallow drier near-sfc layer, and the lowest 100 mb temp profile being less than dry adiabatic... the potential temperature and the mixing ratio of the SB parcel are smaller than that of the ML parcel; and thus the SB parcel path rises to the left of--and accordingly has weaker CAPE and stronger CIN than--the ML parcel.)  Parameters from the sndg are as follows...

Unmodified model sfc T/Td: 62/55 F (thermodynamic parameters below are based on unmodified sndg)
Observed sfc T/Td: 63/55 F
SBCAPE: 803 J/kg
SBCIN: 178 J/kg

Unmodified model ML T/Td: 64/56 F
MLCAPE: 893 J/kg
MLCIN: 145 J/kg

0-3 km MLCAPE: 0 J/kg
MLLCL: 614 m
MLLFC: 2945 m

Bunkers storm motion: 209@ 27 kts (storm-relative parameters below are based on this motion and the unmodified model hodo)
Observed storm motion: 235@15 kts (0-1 km SRH "area" using this motion is swept out for visual purposes on the hodo)
0-1 km SRH: 436 m2/s2
0-3 km SRH: 557 m2/s2
0-6 km bulk shear: 57 kts
5 km sr wind: 18 kts
10 km sr wind: 41 kts
0-1 km MLEHI: 2.4

Tuesday, February 12, 2013

Hattiesburg MS EF-4 tornado environment

I had the day off Sun Feb 10 (and was watching a TWD marathon) and not paying a lot of attention to the weather until after darkfall, when I began to catch photos and video of a fairly large and certainly destructive tornado which had struck Hattiesburg, MS.

A quick review of data indicated a slow-moving upper level trough in place from ND-MT southwesteward to the lower CO river valley.  An occluding/stacked deep layer low would move from wrn NE to srn MN throughout the day.  Along/just ahead of the trailing cold front, a surface-based squall line had already initiated by midnight local time across nearly the entirely length of TX (SPS to DLF) along and just ahead of the trailing cold front... supported by ML dew points of 52-60F and high-level flow that would gradually become very difluent and divergent through the remainder of the day.  By dawn, rather rich moisture well downstream was in position to advect from the wrn Gulf of Mexico into the lower MS river valley... with ML dewpoints of 64-68F, and a moist layer depth of 1.5 to 2.5 km (with the remnant EML having already lifted nearly entirely for surface-based storms soon after).  The threat for severe storms would naturally be enhanced along and just south of the retreating warm front as storms within and ahead of the squall line swept into the destabilizing warm sector.  Some of the more destructive tornado outbreaks with several discrete tornadic storms do seem to be associated with a slow-moving large scale trough and a band of marked difluence/divergence ejecting out to the southeast over a deep moist layer... although in the end, convective mode did become a bit on the muddled side with lots of storms on Feb 10.

As early as 1 PM local time, pre-squall line storms were already plaguing northern portions of the LA "panhandle" and the adjacent southern 1/3 of MS.  The upswing in pre-squall line convection was further supported by a considerably strong corridor of increasing low-level moisture convergence/advection along the gradually veering 50-kt synoptic LLJ axis.  Accordingly, by 22-23Z a number of storms evolved into discrete supercells along and south of the effective warm front.  Interference among various supercells and individual seeding from upstream convection may have kept a more widespread/damaging tornadic event from occurring, but the supercell which tracked ENE through Hattiesburg was one storm which obviously thrived in a somewhat less messy local area on the southernmost flank of the training convection.  The fact no lives were lost is a wonder and a blessing, as this was a EF-4 rated tornado.

The LIX RAOB was launched about the same time the tornado initiated and 60 nm to its south--free of convection.  Although the Hattiesburg area was affected by convection for much of the~5 hours prior to the tornado... during nearly an hour preceding the arrival of the tornado, the inflow environment "cleared out" aside from some possible "sprinkles" from the storm's own anvil business given 8-10 km flow averaging 280 degrees.  However, the surface layer via the ob had remained relatively mild (near 70F) all afternoon considering the preceding thunderstorms... and didn't really climb all that much during the period of pristine inflow--so a moot point, perhaps, esp given the rather rich/deep moist layer. 

Anyway, I modified the lowest ~35 mb of the LIX RAOB very slightly based on obs from KHBG; e.g., the surface temp was reduced ~1F and the surface dew point ~1.5F.  The RAOB surface wind, 21006, was perhaps a bit too light/veered--with 18510 being more appropriate based on the "clear inflow" period as well as obs closest to the general south--but I didn't adjust the hodograph and present it in a different form because the hodograph shape and vertical shear parameters would have changed very, very little.  A final note on the shear environment... the near-surface hodo (wholly around 180-190 degrees in the surface to kink layer, if "modified") was relatively backed considering 900 m AGL flow having veered all the way to 225 degrees (from due SW).  Also, the kink height was quite low for a U.S. significantly tornadic supercell occuring near or east of the MS river: 113 m AGL.  Then again, the most well-defined kinks are not characterized by as much windspeed increase above the kink as this one, so this is not a particularly classic case of a sharp low level hodograph kink (which are most common west of the MS river).  Below are the surface-based and lowest 100 mb mixed-layer thermodynamic parameters from the modified LIX RAOB as well as the vertical shear parameters from its unmodified hodograph (both of whose images appear below).

Sfc T/Td: 71/66.5°F
SBCAPE: 567 J/kg
SBCIN: 73 J/kg

ML T/Td: 75/67°F
MLCAPE: 1083 J/kg
MLCIN: 8 J/kg

0-3 km MLCAPE: 91 J/kg
MLLCL: 566 m
MLLFC: 1419 m

0-1 km SRH: 305 m2/s2
0-3 km SRH: 360 m2/s2
0-6 km bulk shear: 67 kt
storm motion: 245° @ 40 kt

Wednesday, January 30, 2013

Observed environment of killer tornado in Adairsville GA

On January 30, a discrete cell moved northeastward from central AL to nwrn GA... by which time a long-lived QLCS began to "absorb it" from behind.  Storm motion was 50-55 kt during this period as it gradually curved from direction ~210° to ~225°.  Just about the time it was becoming absorbed by the QLCS (~1615Z) the cell spawned a rather substantial-looking tornado in and around the city of Adairsville GA.  This tornado killed at least one person and overturned numerous vehicles per the LSR.  WFO Peachtree City, 55nm to the SSE of the tornadic storm, launched a special RAOB about 45 minutes us a somewhat rare look at the observed environment of a tornado.  A link to a short video clip of the tornado is here:

I modified the RAOB very slightly via cooling the near-surface layer a bit to be perhaps more representative of the inflow; while the surface wind on the RAOB looked sufficiently representative.  Am not getting terribly cute regardless given some showers existed in the inflow environment.

Here are some of the parameters from the modified RAOB.  Should note first, the often typical "weaker SRH" (due usually to a more veered sfc-kink shear vector and/or a weaker LLJ) NAM solution actually was more correct by comparison than the RAP solution was.  Usually the RAP is the winner in this department.  Also, the 2-6 km anticyclonic kink/subtle veer-back-veer pattern with height--and associated SR-flow weakness--fcst by both the NAM and the RAP...which I've found present in a good handful of observed tornadic supercell environments...was not present whatsoever via the RAOB.  i.e., the cell likely had favorable SR flow at all levels.  The angle between the surface to kink shear vector and the storm motion is less than 60 degrees... not atypical of many tornadoes in the ern United States--in fact the hodograph itself is quite reminiscent of some past sig tor events' observed hodographs in MS/AL/TN.  And finally, the RAOB hodograph's "Bunkers right mover" fcst motion for this cell during its tornadic phase nailed the observed motion, with the direction essentially spot on and the speed very close or a touch faster than observed (not surprising given the somewhat shallow convective layer?).  The storm was certainly an EL-buster based on the echo tops/visible satellite presentation.

Sfc T/Td: 68/64°F
SBCAPE: 440 J/kg
SBCIN: 12 J/kg

ML T/Td: 71/63°F
MLCAPE: 428 J/kg
MLCIN: 1 J/kg

0-3 km MLCAPE: 139 J/kg
MLLCL: 566 m
MLLFC: 669 m

0-1 km SRH: 344 m2/s2
0-3 km SRH: 495 m2/s2
0-6 km bulk shear: 75 kt
storm motion: 228° @ 53 kt