26 October 2006 Cold Core Tornadoes over Southwest Kansas   (Case Study)    
by Jon Davies                                                                                                                                                                         

The afternoon of 26 October 2006 saw a number of tornadoes develop over southwest Kansas in association with a cold core 500 mb low aloft.  The setting was a little unusual... the tornadoes occurred farther west than one might expect, with surface analyses suggesting that the main surface low in the pressure field was over north central Oklahoma.  Little if any instability was forecast for southwest Kansas.   Also, most of the tornadoes appeared to be non-mesocyclone / non-supercell in nature (see Mike Umscheid's photos and account).  A case study of the mesoscale setting and environment for this event seems appropriate.

Below is the surface map approaching mid afternoon on 26 October 2006, with low pressure centers and some boundaries indicated.  The main surface low was in north central Oklahoma, with a pacific front acting as an inital dryline and wind shift in central Oklahoma.  As a severe weather forecaster, one's attention might initially be focused here.  But in the wind fields over southwest Kansas, another surface circulation was also visible, probably a secondary low along the WNW-ESE boundary behind the main low.  Notice how this low was closer to the 500 mb low aloft (position also indicated), with a decent westward feed of moisture north of the surface boundary.  
<surface map at 1945 UTC (500mb low position also indicated) 

The SPC surface pressure analysis below at 2000 UTC draws visual focus to the main low in the pressure field over north central Oklahoma.  But, next to it, the SPC analysis of surface vertical vorticity at 2000 UTC (light blue lines) suggests that the secondary low and circulation center over southwest Kansas may be an important feature, with stronger surface vorticity indicated in this area (arrow) close to the approaching midlevel low and cold air aloft.
<SPC surface pressure/wind 2000 UTC  <SPC surface vorticity 2000 UTC

The SPC analyses at 500 mb and 700 mb below show the low aloft moving into southwest Kansas, with cold air at 700 mb (< 0^o C) moving out ahead of the low.
<SPC 500mb analysis 2000 UTC           <SPC 700mb analysis at 2000 UTC

One factor that seemed negative for severe weather was the modest CAPE over southwest Kansas (see SPC analyses of MLCAPE and SBCAPE at 2100 UTC below). At best, the SBCAPE field suggests that there was 200-300 J/kg of CAPE nudging westward into southwest Kansas.  This may not seem like much CAPE, but it can be sufficient for tornado events near closed 500 mb lows.  Surface-based CAPE often works best in analyzing instability for 500 mb cold core events, as mixed-layer CAPE in the lowest 100 mb (see below) tends to average out the CAPE values too much, masking the true buoyancy.
<SPC SBCAPE 2100 UTC                      <SPC MLCAPE 2100 UTC

Looking at low-level thermodynamic fields, low-level CAPE (CAPE below 3 km) is important in 500 mb cold core tornado events as an indicator of developing CAPE that extends close to the ground, a reflection of destabilization from cold air advection aloft and a surface-based environment. The SPC analyses below at 2000 UTC and 2100 UTC showed increasing 0-3 km CAPE in southwest Kansas (> 25 J/kg), not evident prior to 2000 UTC.  Because SPC computes low-level CAPE using a mixed-layer parcel (which tends to average the relevant CAPE too much in cold core events), the actual amount of low-level CAPE was probably greater than indicated.
<SPC 0-3 km MLCAPE 2000 UTC          <SPC 0-3 km MLCAPE 2100 UTC

Surface heating is also very important in 500 mb cold core tornado events.  The zoomed-in satellite photo below over southwest Kansas shows that there was plenty of sunshine south of the WNW-ESE boundary mentioned earlier, increasing surface heating and low-level lapse rates near the boundary, beneath the cold air aloft.
<visible satellite image at 2030 UTC

The surface heating in the sunshine south of the boundary and in the drier air over western Oklahoma showed up as an axis of steep 0-3 km lapse rates on the SPC analysis below (see red dots), pointing northwestward into southwest Kansas near the WNW-ESE boundary.  The LCL analysis next to it also reflects indirectly the surface heating axis, with the axis of highest LCL heights bulging northward into southwest Kansas where LCL heights become dramatically lower near the boundary.
<SPC 0-3km lapse rate 2000 UTC    <SPC MLLCL heights 2000 UTC

The combination of cold air aloft with the approaching 500 mb low, moisture along the WNW-ESE boundary, and surface heating impinging on the boundary where there was low-level CAPE developing, may suggest that a more "unstable" environment was present than evident on the earlier analyses showing meager total CAPE amounts.

Returning to the satellite photo at 2030 UTC (annotated below), the cloud patterns over southwest Kansas were complicated, suggesting a more complex setting of boundaries (dashed black lines) than was seen from the low-resolution surface analysis shown at the start of this case study.  Several boundaries intersections appear to be evident (marked by letters A, B, and C).
<visible satellite image at 2030 UTC, annotated over southwest Kansas to show boundaries and 500 mb low location

As found in Davies and Guyer (2004), boundary intersections within roughly 200 miles of the 500 mb closed low center accompaned by adequate surface dewpoints (low-mid 50s F in this case) and access to sunshine and surface heating are a favored location for tornadoes with cold core 500 mb systems.  From the satellite photo and earlier information, all these features appeared to be present in this case.

By 2100 UTC, low-topped storms developed rapidly at and near boundary intersections A and B in the annotated satellite photo above, and seen in the radar image below at 2122 UTC.  By this time several tornadoes were ongoing in the Minneola area (near boundary intersection B, eastern arrow in radar image below).   Mike Umscheid has excellent photography of some of these tornadoes (click here).   A tornado was also reported on the northwest edge of Ulysees (boundary intersection A, western arrow in radar image below), later damaging the Grant County Hospital.
<lowest elevation base reflectivity from DDC radar at 2122 UTC

Over 25 tornado reports were received in southwest Kansas during the period 2050 UTC to around 2245 UTC, most in the vicinity of boundary intersections indicated in the earlier satellite photo. The last tornado was reported north of Protection with the storm shown on radar below (arrow), and may have been related to boundary intersection C in the satellite photo annotated earlier as the intersection moved east, although this is certainly not clear.
<lowest elevation base reflectivity from DDC radar at 22:45 UTC

A water vapor satellite image at 2115 UTC (below), shortly after tornadoes started being reported, shows the midlevel low centered over extreme southwest Kansas, not far west of where tornadoes were occurring.  A strong dry surge of air aloft is seen in the orange enhancement, with both the Minneola and Ulysees storm clusters beneath this dry "slot' aloft just east of the 500 mb low.
<color-enhanced water vapor satellite image at 2115 UTC

A RUC profile from near Minneola, Kansas at 2100 UTC, modified to reflect the air near the boundary based on blending surface observations from sites such as DDC and P28, is shown below:  With this unusual profile (but not atypical of cold core events), notice that essentially all the CAPE was located below 500 mb, with nearly 1/2 the CAPE located below 3 km !  This would probably translate to rapidly rising low-level parcels in storm updrafts along the boundary. The steep low-level lapse rate (near dry-adiabatic in the lowest 1 km) would also contribute to rapid parcel ascent and stretching in updrafts.
< RUC analysis profile for Minneola KS at 2100 UTC, modified using a 64/54 F surface parcel.

With the afforementioned boundary intersections and pockets of vertical vorticity along the boundaries, it seems likely that the tornadoes on 26 October 2006 were due largely to non-mesocyclone / non-supercell processes. Observations and comments by Mike Umscheid generally confirm this.  It is true that some of the cells on radar or viewed by chasers developed supercell characteristics with shallow mesocyclones and visual RFDs. This may have been related to increaseed horizontal shear with east winds north of the boundaries, as well as strong deep-layer shear in the general environment. But the rapid development of tornadoes was largely confined to boundaries, and, combined with the sounding characteristics above that would strongly enhance low-level stretchng, suggest that this was an event dominated by non-mesocyclone processes. Because thermodynamic characteistics and stratification near 500 mb cold core lows favor enhanced vertical stretching, it may be that many cold core tornadoes events involve non-mesocyclone processes, or those processes combined in some way with supercell tornado processes.

<near Minneola KS   <near Protection KS

A paper I've written coming out in Weather and Forecasting late 2006 or early 2007 discusses tornadoes associated with cold core 500mb lows, and mentions that some events may directly involve non-mesocyclone / non-supercell processes, like this event.

How does one forecast cold core tornado events when the instability is so subtle?  (I didn't exactly see this one coming, myself.)  The small instability certainly makes it hard to pick out settings such as 26 October 2006, so they require some detective work.  Any closed 500 mb low approaching an axis of upper 40s or low-mid 50s F surface dewpoints should be suspect.  Assess surface features such as low pressure centers or surface map wind circulation centers, and, of course, boundaries.  Note boundaries and boundary intersections with adjacent surface moisture and sunshine / heating within 100-200 miles of the approaching midlevel low.   Don't rely on model-based CAPE progs, which usually don't output CAPE less than 500 J/kg.  Surface-based CAPE of roughly 200 J/kg or more can be significant in cold core tornado events!

Below are some morning RUC forecast panels (from the UCAR site and Earl Barker's site) valid for mid afternoon on 26 October 2006:
<RUC 500 mb forecast                   <RUC total MLCAPE forecast

<RUC 0-3 km MLCAPE forecast  <RUC MLLCL forecast

It's subtle, but ahead of the 500 mb low, one can see an area of low-level (0-3 km) CAPE over the Oklahoma panhandle, even though there is _no_ CAPE shown on the total MLCAPE forecast (the display doesn't show values < 500 J/kg).  It turns out that the RUC model put the location of this low-level CAPE a little south of where it actually ended up being. (southwest Kansas). But this should be a tip that there may be enough CAPE near the midlevel low to support possible tornadoes. Next, looking at the MLLCL forecast, it defintiely shows low LCL heights (yellow, meaning low cloud bases) over Kansas, and high LCL heights (blue and white) over the Panhandle area, suggesting an E-W boundary between the two in the area where the low-level CAPE is progged. Also important, notice the bulge northward of high (white) LCL heights forecast over western Oklahoma and the eastern Panhandle area, pointing northward toward the boundary and area of low-level CAPE.  This suggests significant surface heating impinging on the boundary to the north in an area with low-level CAPE and increasing cold air aloft. These are prime ingredients for cold core tornadoes!  These hints suggest careful monitoring of surface, satellite, and environment data over the Panhandle area and southwest Kansas as the day progresses, looking for boundary intersections and surface heating relative to the position of the advancing low aloft.

The above clues are something I wish I had picked up better on 26 October 2006... there's certainly always more to learn  :-).

-  Jon Davies 10/28/06

to Jon Davies main page