Severe Convective Weather: Any Overview of Severe

advertisement
Severe Convective Weather: An Overview of Severe Convective
Storms
By
Jim Bishop
July 21, 2004
I. Introduction
There is currently a broad understanding of severe storms. From the single-cell air mass
thunderstorm, to the supercell thunderstorm meteorologists have been studying severe
storm types and their interaction with one another, the environment, and their own
evolution between storm types for decades. While the supercell is sometimes thought of
as the most complex of the thunderstorm spectrum, a study of it alone is not enough to
encompass the severe convective spectrum. For the number of years thunderstorms have
been studied, it seems we should know a little more about them that we actually due. The
problem lies in our data. In that, the main limiting factor in studying severe storms is the
sparse upper air observations that are currently available across the United States.
Observations from Foote (1977) indicate that there is enough overlap and variation in
severe storm types to assume a continuous spectrum of possible storm structures.
Personal observations and research indicate that storm structure and evolution are highly
dependent upon their wind shear environment. Bluestein and Jain (1985) and Bluestein
et al. (1987a), have found observational evidence that the evolution of convective storms
depends on their environment as well. While it is clear that wind vertical wind shear
plays an important role in storm type, it is not entirely clear the magnitude of the wind
shear that is needed, or the depth of that wind shear that is necessary for a particular
storm type.
Supercell thunderstorms are the most severe and destructive of the thunderstorm
spectrum. They are capable of producing very large hail, damaging straight line winds,
and violent tornadoes. The shear profiles which support supercells are not clear-cut, nor
are they well understood. Part of this paper will attempt to describe wind shear profiles
which would likely support supercell thunderstorms, and will also attempt to further
describe the vertical wind shear which would support tornadic supercells. In doing this,
an understanding of mesocyclones is needed. This becomes somewhat ambiguous as
there is no universally accepted definition of a mesocyclone (Moller et. al., 1994).
The purpose of this paper is to provide an overview of severe storms and the
environments in which they often develop. Contained within are three case studies of
tornado outbreaks from May 12, 2004 in southern Kansas, May 22, 2004 across
Nebraska, Kansas and southwestern Iowa, May 29, 2004 across Oklahoma, Kansas,
Nebraska, and South Dakota, and one derecho event on June 2, 2004 across Kansas,
Oklahoma, Arkansas, Texas, Louisiana, Mississippi and Alabama of 2004. May 22 and
29 were major tornado outbreaks consisting of more than fifty tornadoes, May 12 is a
small outbreak from one incredible supercell, and the last case is a derecho event. Since
three out of four of these cases are tornadic supercell events, and the distinguishing
elements between tornadic and non-tornadic supercells is still not well understood, this
paper will focus more on the conditions necessary for such storms, while also discussing
the mechanisms which encompass damaging straight line wind events.
II. Literature Review
Over the past 30 years, severe storms have been classified according to their degree of
severity, longevity, mode of propagation, intensity of rainfall and other related features
(Weisman and Klemp, 1982). The development of high-resolution Doppler radar, visual
observations from storm intercept teams, and detailed numerical models have increased
our ability to make a distinction between tornadic and non-tornadic storm types (Doswell
and Burges, 1993). Initial tornado forecast studies resulted in empirical forecast rules
that focused on the role of highly baroclinic, synoptic-scale systems (Miller 1972).
However, it has become well known that such easily recognized disturbances are not
always present with supercell storms (Maddox and Doswell 1982).
While there are several different parameters that play a role in storm type, there are two
that seem to be the most significant. As indicated by (Weisman and Klemp, 1982), the
spectrum of convective storm types may be most dependent upon vertical wind shear and
stability (CAPE). Observations from Chisolm and Renick (1972); Fankhauser and Mohr
(1977); and Burgess and Curran (1985) show that vertical wind shear is the most crucial
out of the two parameters for supercell development.
Unorganized convective storms, or airmass thunderstorms, are associated weak wind
shear environments (Moller et. al., 1994). But they also have some CAPE in the local
environment. We could refer to weak shear as having less than 15m/s (29 knots) of 0-6
km shear (Weisman and Klemp, 1982). However, it is again difficult to draw a fine line
between weak shear and marginal shear for supercells. Davies and Johns (1993) found,
through numerical simulations, that the mean 3-6 km winds for supercells was 11m/s (20
knots), which agrees with Doswell’s observations that middle-level winds (winds
between 700mb and 500mb) should generally be 11m/s (20 knots) or greater regarding
hodographs that indicate potential for supercell development. Johns (1983) noted that the
many variations in the form of the environmental wind shear make it difficult to evaluate
the general magnitude by visual inspection.
The multicell thunderstorm is the first type in the thunderstorm spectrum which is
typically severe. It only requires weak vertical wind shear. According to a study done by
Weisman and Klemp (1982), 15m/s (29 knots) of 0-6 km shear is needed to support
multicell thunderstorms. However, I due believe those results are heavily biased to
stronger wind magnitudes, which may not necessarily be needed in every situation.
Multicell thunderstorms have a relatively short life span as new updrafts are continuously
forming along the old one’s outflow. For that reason, it makes sense that multicellur
storms do not need strong deep layer wind shear to support them as the processes that
keeps the “system” going is the cold outflow from the rain near the updrafts. One of the
effects of strong storm-relative winds, or strong deep layer shear, is to blow rain away
from the updraft (Brooks et al., 1993). Appropriately, the storm-relative winds associated
with most multicell storms are not strong, owing to their intense outflow and new updraft
generation every several minutes. This is likely a significant factor in determining
whether a storm will be multicellular or supercellular.
Linear Storm Modes
Squall lines and bow echoes come next in the severe thunderstorm spectrum. Fujita
(1978) introduced the term “bow echo” in the late 1970’s to describe bow-shaped radar
reflectivity signatures associated with the majority of straight line wind storms, which are
noted for producing long swaths of damaging straight line surface winds (Doswell III and
Evans, 2001; Weisman, 1992). Both Fujita and, more recently, Przybyliski and Gery
(1983) emphasize the association between bow echoes and long swaths of damaging
straight-line winds (Weisman, 1992). Squall lines and bow echoes are associated with
relatively strong vertical wind shear. Long-lived bow echoes are the most damaging, and
are referred to as derechos. The term “derecho” is used to describe convective systems
that produce straight-line convective wind gusts great than 26 m/s (50 knots) within a
concentrated area with a major axis length of at least 400 km (Weisman, 1992).
Derechos are associated with bow echoes or bowing segment of varying scales within a
squall line (Przybylinski and Decaire 1985, JH87). There are two types of derechos. One
evolves from rapidly propagating segments of intense squall lines, and is often associated
with strong low pressure systems. The other is most common in the late spring and early
summer when the flow aloft is northwesterly (Johns and Hirt, 1986).
A study conducted by Johns et al. (1990) from 14 intense derecho events from June and
July of 1990 revealed the average 0-6 km shear was 20 m/s (38 knots). However,
Doswell III and Evans (2001) indicated that most of these events also developed and
persisted with 0-6km shear less than 20 m/s. Looking back to Weisman and Klemp
(1982), suggesting 15 m/s will support multicell thunderstorms; there is a distinct
correlation between multicell thunderstorm shear environments and squall line/derecho
wind shear environments. Interestingly, Thorpe et al. (1982) noted more specifically that
simulated squall lines were strongest for environments in which the vertical wind shear
was confined to low levels. While there is general agreement on the importance of low
level shear, there is much less agreement on why it is important (Weisman and Rotunno,
2003). A case study, conducted by Weisman (1992) shows a unidirectional vertical wind
shear of 25 m/s (~49 knots) over 2.5 km in an environment where a squall line, oriented
perpendicular to the shear vector, existed. However, Weisman does not mention what the
initial storm mode was in the early stages of convective development. Johns and Hirt
(1986) noted that the average 700 mb wind speeds were 17 m/s (34 knots) for several
derecho events. Based on these findings, it is evident that there is more to the puzzle than
simply vertical winds shear. Forcing, both frontogenetic and large scale, likely plays a
role in the initial storm type.
Squall lines, bow echoes, and derechos are all related in that bow echoes are a smaller
scale, more intense phenomena associated with squall lines, and derechos are simply
strong, long lasting bow echoes, as defined above. Therefore, an examination of
thermodynamic and synoptic conditions for any or all of these phenomena will provide
clues as for the development of any or all three, generally speaking.
The system usually begins as a single, large and strong convective cell that may be
relatively isolated, or may be part of a more extensive squall line (Weisman, 1992). For
the case of Weisman (1992), the CAPE was observed to be 2400 J/KG. Doswell III and
Evans (2001) examined a case where the CAPE increased to an average maximum of
4500 J/KG as the convective system moved east. Johns and Hirt (1986) examined 70
derecho cases to reveal extreme convective instability was present along the derecho
tracks. Surface dewpoints greater than 20°C are common, with lifted indices
approximately -9°C (Weisman, 1992). Furthermore, it is evident that copious amounts of
low-level moisture are present during derecho events, along with extreme instability.
Synoptically, it seems (from 70 derecho cases) that most derecho events, or events with
widespread damaging straight line winds, are associated with a westerly or northwesterly
flow at 500 mb (Johns and Hirt, 1986). They are almost always associated with a quasistationary, east-west oriented thermal boundary near the surface, which is parallel to the
mean flow. This boundary provides a thermal gradient across which warm air advection
develops (Johns and Hirt, 1986). This war air advection regime usually shifts eastward
with the convective system along the thermal gradient. The pooling of moisture in the
convergence zone along the boundary is common, owing to copious amounts of low-level
moisture.
Cold outflow from downbursts is the cause for most of the straight line wind damage in
bow echoes and derechos. According to Johns and Hirt (1986), there are two primary
contributors to outflow intensity. The first is the negative buoyancy, which is created by
evaporation of precipitation. The second is the transfer of higher momentum air aloft to
the boundary layer (Newton, 1950; Fujita, 1959). The first implies drier air entrainment
in the middle levels (3-6 km AGL), which would allow for optimum evaporation of
precipitation. Johns and Hirt (1986) found that most of the significant entrainment of
environmental flow into the downdraft occurs in the mid-troposphere. Based on those
finds, they concluded that derecho development would be aided by relatively strong
winds and low humidities in the 3 to 7 km layer. Browning and Ludlam (1962); and
Hooking (1965) also mention evaporative cooling when precipitation falls through a layer
of unsaturated air as an important factor in the development and maintenance of strong
downdrafts. Doswell III and Evans (2001) support the latter contributor to outflow
intensity, stating that the strength of the mean flow, and its possible effects on the speed
of momentum, enhance the development of sustained severe wind gusts at the surface.
Droegemeier and Wilhelmson (1985) found through numerical model simulations that the
speed of convective downdrafts becomes greater as the low-level moisture is increased.
This ties in nicely with the incredible amount of low-level moisture that is present during
derecho events. Furthermore, recent numerical simulations by Srivastava (1985) suggest
that the lower densities associated with the high relative humidities in the subcloud layer
may enhance the negative buoyancy of the downdraft. These are two theories which may
enhance the effects of the rear-inflow jet, which cause most of the damaging straight line
winds within bow echoes. Fujita proposed this to be the source of damaging winds in
bow echoes, which is thought to be associated with the bulging and speeding up of the
radar echoes (Weisman, 1992).
Supercell Thunderstorm Modes
Supercell thunderstorms require strong, 0-6 km shear, as noted by Davies and Johns
(1993) that sufficient values of the wind parameters must be present for thunderstorms to
develop into supercells that produce tornadoes. Through personal observation, the actual
value of shear through the lowest six kilometers seems to not only be subjective, but also
dependent upon the initiating factor and the amount of forcing involved. For the
numerical simulations of Weisman and Klemp (1982, 1984), 20-25 m/s of 0-6 km shear
corresponds to approximately the minimum shear necessary to produce storm splitting/a
mesocyclone, or a supercell thunderstorm, given CAPE of about 2200 J/KG. Since this is
based on a numerical simulation, and I have seen supercells with less 0-6 km shear
present in the environment, I am forced to assume some error in this model, as the
atmosphere cannot be perfectly duplicated. Davies and Johns (1993) stated that the
weakest 0-6 km mean wind speeds found in a 31 case subset of supercell days were near
10.3 m/s (20 knots), suggesting that mean wind speeds of lesser magnitude may be too
weak to support strong or violent tornadoes in supercell thunderstorms. Davies and Johns
(1993) also defines relatively “weak” wind fields that can support significant supercellinduced tornadoes as those with 0-6 km mean wind speeds less than or equal to 15.5 m/s,
or 30 knots and greater than or equal to 10.3 m/s (20 knots). Based on these findings it
seems a 0-6 km shear range of 20-30 knots is the minimum wind shear criteria to support
supercell thunderstorms.
While the 0-6 km shear (referring to speed shear) is the most important vertical wind
shear parameter, directional shear has significance as well, and is often present during
severe weather outbreaks. In 1963, Newton stated from forecasting experience that
severe local storms are generally associated with winds which not only increase with
height but also veer with height (Browning, 1964). According to Johns (1983) the
directional contribution to environmental wind shear is a major factor in providing
sufficient storm relative inflow for severe storm maintenance. From personal forecasting
experience, it seems approximately 45° of directional shear is needed in most severe
weather events. Furthermore, most, but not all, severe weather outbreaks are associated
with significant (45 degrees or greater) veering from the surface to 6 km. Rotunno and
Klemp (1982) found that the sense of the directional shear determines where supercell
storms develop (Schaefer, 1986). Lilly (1982, 1983) demonstrated that long lasting
storms would require an environment in which the product of the wind speed and the
variation of wind direction with height is large, while Davies-Jones (1983) showed that a
strong veering of the winds across the subcloud layer should favor mesocyclone
formation. From these finds it seems apparent that a strong veering wind profile in the
lower levels (0-3km) with strong speed shear over 0-6km is indicative of supercell
environments. In situations with strong 0-6 km shear, the storm motion will be such that
the storm relative winds veer considerably with height; owing to significant directional
shear. Depending upon other factors such as frontogenetic forcing and strong upper level
support, stronger shear is likely necessary in situations when such stronger forcing is
involved. It should be noted that under circumstances in which the initiating boundary
and/or forcing is perpendicular to the shear vector, directional shear is not needed to
support supercells if adequate speed shear is present.
Supercells are relatively rare when looking at the overall severe storm spectrum. Even
during the spring months in the American Great Plains only a small percentage of
thunderstorms contain mesocyclones (Moller et al., 1994). This brings me to the
definition of a supercell, which according to Moller et al. (1994) is a convective storm
that contains a mesocyclone throughout a significant portion of the storm’s life. Based
on that definition, there is actually no clear-cut definition of a supercell, since there is no
clear-cut length of time in which a storm must contain a mesocyclone. Thus, we again
are discussing something which is rather subjective. We need to know what defines a
mesocyclone, and Moller et al., (1994) attempts to do so. Moller et al. (1994) says that
mesocyclones are assumed to have vertical vorticity greater than or equal to .1/s,
lifetimes at least on the order of tens of minutes and are present through a substantial
fraction (>1/3) of the convective storm’s depth. So a mesocyclone is a strong, persistent,
rotating updraft that exists throughout a great depth in the storm.
While certain types of severe storms require certain magnitudes of wind shear, all severe
thunderstorms require moisture, CAPE, and some lifting mechanism to trigger them. In
1954, Fawbush and Miller (1954a) published a list of three types of severe weather
soundings. One of them stresses a need for a deep layer of warm, moist, conditionally
unstable air. A typical morning severe weather sounding features a surface-based moist
layer averaging 175 mb thick that is capped by an inversion about 45 mb thick (Schaefer,
1986). So it is understood that a moist profile from the surface to near the 850 mb
pressure level is a necessity for severe convection to occur. Figure 1 is a good example
of a typical morning severe weather sounding.
There is a natural question concerning the role of CAPE in supercells. Moller et al.
(1994) could not definitively state the lowest value of CAPE that is still supportive of a
supercell. It is only known that “some” CAPE is necessary for any convective storm, but
the exact amount has yet to be determined. However, the absence of high CAPE does not
exclude the possibility of supercells. According to Korotky (1990), damaging supercells
do occur with low CAPE (e.g. the northern Indiana tornadoes of 3 April 1974 and the
Raleigh, North Carolina tornado of 28 November 1988). Even more recently there have
been damaging supercells containing violent tornadoes with a low CAPE environment.
One such event occurred on 10 November, 2002 across northern Ohio. Johns and
Doswell (1992) noted that a substantial fraction of supercells nationwide are in
environments with CAPE less than 1500 J/KG. In a supercell study done by Weisman
and Rotunno (1999), a model was run using 2200 J/KG of CAPE, which worked well for
their model. So, it might be accurate to state that 1500-2500 J/KG of CAPE will support
supercells. Whether or not this is the ideal amount of CAPE for supercells is not only
subjective, but the ideal amount may be different for shear environment0.
Upper Level Support
Upper level support (i.e. vorticity advection increasing with height, jet streak coupling,
ect.) is not necessary for severe convection to occur. However, most severe weather
events have an associated capping inversion in the middle levels of the atmosphere,
preventing convection from occurring until the combination of upper level support and a
triggering mechanism can break the cap. The most common form of upper level support
is positive (cyclonic) vorticity advection increasing with height. Most of the time this
can be seen looking at a 500 mb height chart and noting where the greatest speed shear
and curvature are located within a mid-tropospheric shortwave. According to Doswell
(1982) when the 500 mb height contour intersects the vorticity pattern at > 30 degrees,
there is strong vorticity advection in that region. Thus, if we assume little or no vorticity
near the surface, then we can assume vorticity is increasing with height up to the 500 mb
level. Thus, there is vorticity advection increasing with height, owing to upward vertical
motion throughout the column from the surface to 500 mb. 500 mb vorticity advection
increasing with height is often a significant feature, cooling the midlevels, which helps
weaken the cap, allowing a triggering mechanism to initiate tornadic supercells during
major tornado outbreaks.
There are often times very strong winds aloft associated with strong mid-latitude
cyclones. Branching from this, severe thunderstorm forecasters have noted that there is
sometimes a relationship between isotach maxima in the jet stream, jet streaks, and
severe weather activity (e.g., Beeb and Bates, 1955; Danielsen, 1974). There are several
possible explanations for these occurrences, most of which depend upon the existence of
transverse vertical circulations. On the right entrance and left exit regions of a jet streak,
divergence is due to a downstream increase in the ageostrophic wind speed, while
convergence is a result of a downstream decrease in the ageostrophic wind speed on the
left entrance and right exit regions. If the jet streak is near the tropopause where vertical
motions are negligible, if there is little or no temperature advection at any level, and if
temperature decreases to the north, then there is rising motion on the left exit and right
entrance regions of the jet streak (Bluestein and Thomas, 1984). An example is shown in
figure 3 (although this is not a good example of strong divergence associated with the jet
streak). The rising motion can be coupled with moisture convergence at low-levels,
thereby helping to maintain a region of towering cumulus, or even help break a capping
inversion. Divergence can also decrease the static stability below that level (Ucellini and
Johnson, 1979).
Initiating Mechanism
Most of the classic tornado outbreaks that have occurred in the United States have been
associated with a strong baroclinic system. In the Great Plains during the spring time a
dryline forms in association with the baroclinic system. The typical triggering
mechanism for most of the Great Plains tornado outbreaks is the dryline (Rhea, 1966). In
1982, Doswell stated that a key element in producing a severe weather forecast is the
identification and prognosis of surface boundaries. It is certainly true that while upper
level support will weaken the capping inversion, which in turn helps to initiate
convection, surface boundaries (e.g. drylines, stationary fronts, warm fronts, and outflow
boundaries from previous storms) are usually the final triggering mechanism for
convective storms (Moller et al., 1994). A good example of a dryline is shown in Figure
1 when supercells were initiated across northwestern Oklahoma on May 15, 1991
(Bluestein et al., 1993).
Orography plays an important role in the initiation of severe convection; especially across
the high plains of eastern Colorado, eastern New Mexico and the Oklahoma and Texas
panhandles. Synoptic scale cyclogenesis does not occur randomly or only due to
vorticity advection or jet streak coupling. Rather, it is concentrated in certain key
geographic areas, often times in the lee of major mountainous regions (Roeber 1984;
Tibaldi et al. 1990). The most common place for lee cyclogenesis in the Great Plains is
eastern Colorado, as a result of the Rocky Mountains.
Gradual lift from upslope flow also plays an important role in the initiation of severe
convection. The day before a strong baroclinic system ejects into the southern or central
plains, there are strong pressure falls over Colorado and/or New Mexico, resulting in
strong surface winds from the southeast or east over the high plains of the Texas
panhandle, Oklahoma panhandle, and eastern Colorado. A parcel’s gradual increase in
elevation can and usually does initiates convection. There is usually sufficient 0-6 km
shear to support supercells in these setups.
Vertical Profile
As mentioned earlier, a deep layer of warm, moist air in the lowest three kilometers is
optimal for supercell thunderstorms. Figure 1 is an example of the classic FawbushMiller “loaded gun” sounding. This sounding has a deep moisture profile from the
surface to 850 mb. It has a pronounced capping inversion between 850 mb and 800 mb,
and an associated elevated mixed layer of dry air just above it. The sounding also has
another elevated mixed layer near the 500 mb level. While this sounding was launched in
the early afternoon, weakening of the cap and modification of the air mass was still
occurring. As noted by Carr (1952), during the day the Fawbush-Miller “loaded gun”
sounding is modified so by mid-afternoon the low-level moist layer is quite thick and the
inversion is significantly distorted (Schaefer, 1986). This occurs due to the warm air
advection at low levels which advects higher dewpoint temperatures into the region.
Daytime heating and midlevel cooling due to vertical velocities associated with upper
level features are two of the most significant parameters which cause the capping
inversion to weaken during the afternoon. This is supported by Aufm Kampe (1960) who
proposed that a mountain wave type disturbance in the flow create variations in both the
strength of the capping inversion and in the lapse rate of the air above it (Schaefer, 1986).
However, Danielson (1975) showed that the forced subsidence in the midlevel flow
crossing the Rocky Mountains can give rise to the warm, dry air mass which is positioned
above the low-level inversion of the “loaded gun” sounding (Schaefer, 1986). This will
actually strengthen the capping inversion. Therefore, we must consider both the
parameters which can weaken the cap and those which can strengthen the cap, and decide
which one is more significant. Only then will we know the true strength of the cap.
There is still much to be understood about the environments associated with the spectrum
of mesocyclone storms (Moller et al., 1994). What is understood and well documented
are the three basic types of supercell thunderstorms: the classic (CL), high precipitation
(HP), and low precipitation (LP) supercell. Each type of supercell forms in slightly
different shear and moisture profiles.
Supercell Types
Classic supercells (CL) are usually isolated, developing well apart from competing
storms (Moller et al., 1994). They are characterized by upshear flanking convective lines
with rain free bases and a well defined lowering/wall cloud from which tornadoes
sometimes form (Figure 3a). Visual observations indicated that rain typically wraps
around the back side of the updraft, while still allowing for visual sighting of the wall
cloud and tornado. On radar this feature creates a “hook” structure which is one of the
most common radar signatures of a classic supercell. Another common radar signature is
the Bounded Weak Echo Region (BWER), which indicated where precipitation is being
suspended in the air due to an intense updraft (Chisolm and Renich 1972; Lemon 1980).
According to (Penn et al. 1955; Fujita et al 1970; and Forbes 1981), as well as evidenced
in Environmental Science Service Administration/National Oceanic and Atmospheric
disaster survey reports, is that major tornado outbreaks are dominated in numbers by
classic supercells. The reasoning for this finding is part of the purpose of this paper. It
seems when the shear profiles are optimum for supercell thunderstorms, and the moisture
profiles are also optimum, classic supercells generally form.
Low precipitation supercells (LP) are the least understood of the supercell spectrum.
This is mainly due to the fact that these storms are known for producing very large hail,
yet it is still not well understood why this occurs. These storms do not have the same
radar signatures as classic supercells but are visually spectacular in revealing rotation
(Bluestein and Parks 1983). They can appear benign on radar, often exhibiting low
reflectivities despite producing very large hail. Severe weather associated with them is
usually limited to large hail and only occasional tornadoes. They are characterized by
low-to-moderate moisture values and relatively high LFCs. LP supercells are virtually
unique to the dryline environment in the high plains region east of the Rocky Mountains
and western portions of the Great Plains. No violent tornadoes have been observed with
an LP supercell. The lack of any significant rain precludes much chance for damaging
outflow winds and flash flooding (Moller et al., 1994). See Figure 3b.
High precipitation (HP) supercells are characterized by substantial precipitation in and
near the mesocyclone (Figure 3c). Visual observation of tornadoes that form from HP
supercells is difficult since rain almost always obscures the tornado. According to Johns
et al (1993), the HP supercell is the dominant supercell nationwide. Interestingly, a
significant fraction of the classic supercells in outbreaks appear to have been HP
supercells (some may have undergone transition from CL to HP) (Moller et al., 1994).
Increased low level, vertical wind shear appears to be particularly important for many
HP-type supercells (Moller et al. 1990). It has been found that weaker winds aloft and
high precipitable water values are often associated with HP supercells.
It is rare for a supercell to exhibit al the criteria for a specific “type” of supercell.
However, the presence of a persistent “cell” is the most commonly accepted radar
characteristic associated with supercells (Doswell and Burgess, 1993). With the single
cell comes the notion of a steady state character. It is unlikely that supercells have a
steady state updraft, since many go through cycles, constantly evolving in cellular
structure. Doswell and Burgess suggest that supercells do have a “background process”
that evolves slowly over periods of a few hours, which is the characteristic lifetime of the
constantly regenerating supercell structure.
Mesocyclones
A mesocyclone is a rotating updraft which is assumed to have vertical vorticity greater
than or equal to .01/s, and last at least on the order of tens of minutes, and is present
through a substantial fraction (>1/3) of the convective storm’s depth (Moller et al., 1994).
It is well understood how the midlevel mesocyclone develops in supercell thunderstorms.
It is the result of environmental shear creating vorticity. This vorticity is stretched and
tilted into the vertical, resulting in a vortex pair straddling the updraft (Rotunno).
Subsequent storm splitting implies a decreased vertical velocity, and eventually a
downdraft, which develops in the location of the original updraft. There is now one
cyclonically rotating updraft (right split) and one anti-cyclonically rotating updraft (left
split), each containing a downdraft as well. When we refer to the midlevel mesocyclone,
we are referring to the cyclonically rotating updraft. This theory was first illustrated by
Klemp in 1987. The idea works great for midlevel rotation, but still does not explain the
formation of a low-level mesocyclone, which is needed for violent tornadoes. But,
Brooks et al. (1994) found through both numerical modeling simulations and
observations, that convection typically does not develop low-level mesocyclones in the
absence of an existing midlevel mesocyclone.
The origin of low-level mesocyclones is still undergoing considerable research. The
reason for studying the low-level mesocyclone is this is what most significant tornadoes
are believed to form from. The problem is that half of all mesocyclones produce
tornadoes (Brooks et al., 1993). Brooks et al. (1993) shows examples of what are
referred to as “failure modes”, in which storms with midlevel mesocyclones fail to
produce low-level mesocyclones, hence, would be unlikely to produce significant
tornadoes (Brooks et al., 1993). Appropriately, Brooks et al. (1994) found evidence that
low-level mesocyclones develop by different processes than midlevel mesocyclones. In
order to study low-level mesocyclones, one would think of studying the low-level wind
fields in the local environment. However Rasmussen and Wilhelmson (1983) state that
forecasting techniques that deal only with the low-level wind field of the environment,
such as low-level shear magnitude, are essentially forecasting supercells, not tornadoes
(Brooks et al., 1993). I find this misleading considering that deep layer shear is necessary
for supercells. Once you have that, then a low level mesocyclone may form due to a
certain magnitude of low-level shear. Nonetheless, this is merely speculation on my part
from field observations. Rasmussen and Wilhelmson’s theory is reinforced by
Wilhelmson and Klemp (1978), who state that low-level shear (0-3km) is more important
to the development of modeled supercell structures that upper level shear (Weisman and
Klemp, 1982).
Through several years of supercell observations, it appears that evaporatively cooled air
from the rear-flank downdraft, combined with a baroclinic process which is entrained
into the updraft, may help to form or enhance the low-level mesocyclone. Davies Jones
and Brooks (1993) found in their model that the presence of evaporating precipitation
near the updraft in the rear-flank downdraft is essential to the development of low-level
mesocyclones (Brooks et al., 1993). Rotunno and Klemp (1985) demonstrated that the
origin of the low-level mesocyclone depends on baroclinic processes resulting from the
evaporation of rain (Brooks et al., 1993). Davies-Jones and Brooks (1993) then extended
their argument to show that tilting of baroclinically generated horizontal vorticity within
the rear-flank downdraft could produce positive vertical vorticity within the downdraft.
As a result, air reaching the ground within the downdraft could have positive vertical
vorticity before it encounters the updraft (Brooks et al., 1993). So, it is clear that there is
some interaction between evaporatively cooled air in the rear-flank downdraft and the
updraft, involving baroclinically produced vorticity that is significant in the development
of a low-level mesocyclone.
Based on these findings from numerical simulations, it seems that low level shear (0-3
km) is critical for supercell development. Therefore, increased low level, vertical wind
shear from locally backed surface winds along thermal boundaries (e.g. warm fronts,
outflow boundaries, etc.) can aid supercell development in most cases (Maddox et al.
1980; Moller et al., 1994).
Since it is clear low-level shear (0-3 km) is the most critical vertical wind depth for
supercell or midlevel mesocyclone development, a question arises about deep layer shear
(0-6 km). Is deep layer shear really that important for supercell development (which is
needed to support a low-level mesocyclone)? Observationally, most early work on the
environments around supercell thunderstorms and tornadoes stressed the need for strong
mid and upper-level winds (e.g., Fawbush et al. 1951); (Shuman and Carstensen 1952;
Fawbush and Miller 1952, 1954; Brooks et al., 1993). For clarification, Doswell (1991)
specifies middle-level winds as those in roughly the 700 mb to 500 mb layer (Davies and
Johns, 1993). One of the effects of the strong midlevel storm-relative winds is to blow
rain away from the updraft (Brooks et al., 1993). Davies and Johns (1993) found that
strong middle level winds may move precipitation downwind out of the upper portion of
the updraft, thereby eliminating a potential impediment to the development of a strong
and sustained updraft; or a strong mesocyclone. This makes sense because if strong 0-6
km shear is present, then there must be at least adequate winds aloft which will blow
precipitation away from the updraft. This allows the updraft to sustain itself without the
influence of rain or downdraft air affecting it. This supports the current knowledge of
supercell shear environment in which strong deep layer shear is necessary to support
supercells.
This ties in nicely with low-level mesocyclone development. According to (Brooks et al.,
1994), for a given mid-level mesocyclone circulation, intensifying the midlevel stormrelative winds increases the amount of rain blown away from the updraft, thereby
lessoning the low-level baroclinic generation and the development of strong outflow.
Based on that, it appears as if a necessary condition for the development of a supercell,
and a sustained low-level mesocyclone, for the storm relative inflow to be strong enough
to keep the outflow from propagating away from the updraft (Weisman and Klemp,
1982). If outflow propagates too far away from the updraft, it essentially undercuts the
low-level mesocyclone and destroys the tornado potential for a period of time. In the
case of weak storm-relative winds, any low-level mesocyclone which occurs early in the
storm’s life will be short lived, with the outflow dominating the storm. For strong stormrelative winds, baroclinic generation will be small, but then again so will outflow. As a
result, a low-level mesocyclone might be very slow to develop or perhaps might not
develop at all, but the outflow will be relatively weak (Brooks et al., 1994).
Subsequently, if a low-level mesocyclone does develop, it is much more likely to last for
an extended period of time. So there must be some balance between the midlevel winds
and precipitable water content, allowing for optimum low-level mesocyclone formation,
without any outflow disrupting the process.
Weisman and Klemp (1982) found that the existence of a strong low-level mesocyclone
(vertical vorticity maximum) is closely related observationally to the development of
tornadoes. More recent studies conducted by Wicker and Wilhelmson (1995) expand on
Rotunno and Klemp’s (1985) idea that the source of vorticity in tornadoes associated
with mesocyclones in supercells is thought to be horizontal vorticity at low levels, either
produced locally through baroclinic processes or imported from elsewhere; the vorticity
is tilted into the vertical and subsequently amplified by stretching underneath an updraft,
driven in part by buoyancy and in part by an upward-directed perturbation pressure
gradient force (Bluestein et al., 2003).
When discussing the overall severe convective storm spectrum, there is still a great deal
to be understood. The evolution of long-lived bow echoes and derechos has been
documented, but a clear understanding of them is still undergoing research. Digging
deeper into the spectrum, the point has been established that, generally speaking,
supercells are relatively well understood, while tornadic supercells are not. This refers to
the current understanding of the lowest 3 km of the atmosphere where the low-level
mesocyclone develops. Until a vast array of radiosonde data sets is established near
tornadic supercells, the lowest 3 km may remain ambiguous.
Figure 1: Surface analysis of a southern plains dryline from May 15, 1991. Supercells initiated off of the
dryline across northwestern Oklahoma later in the afternoon (Bluestein et al., 1993).
Figure 2: KBMX 1800 UTC Birmingham, Alabama sounding from May 5, 2003. This is an excellent
example of the classic “loaded gun” sounding, which characterizes an atmosphere which is capped, yet
potentially capable of supporting severe storms.
Figure 3: The 250 mb analysis from March 25, 2002. The divergent regions of a jet streak located over
the Great Lakes have been circled. Keep in mind this is not a great example of strong divergence, but is
being used as a visual aid.
Figure 3a: Two perspective of a classic (CL) supercell. Left: Low level radar structure and cloud features
looking down from above. Right: Visual structures from a nearby observer on the ground.
Figure 3b: Two perspective of a low precipitation (LP) supercell. Left: Low level radar structure and
cloud features looking down from above. Right: Visual structures from a nearby observer on the ground.
Figure 3c: Two perspective of a high precipitation (HP) supercell. Left: Low level radar structure and
cloud features looking down from above. Right: Visual structures from a nearby observer on the ground.
III. Case Studies
The following are four severe weather case studies, consisting of three tornado outbreaks,
two of which were significant, and one derecho event: May 12, 2004 southern Kansas
tornado outbreak; May 22, 2004 central plains tornado outbreak; May 29, 2004 central
and southern plains tornado outbreak; and June 2, 2004 southern plains and Gulf coast
derecho.
Case I: May 12, 2004 southern Kansas tornado outbreak
On May 12, 2004 several supercell thunderstorms occurred in southern central Kansas
and parts of western Oklahoma. At least two of the supercells in southern Kansas were
tornadic, with a total of 13 tornadoes which occurred in Barber and Harper counties of
southern Kansas (Figure 3b). The largest of the tornadoes near Harper, Kansas in Harper
County was rated an F4. Large hail and damaging winds were the only severe reports
received from the supercells across western Oklahoma. The purpose of this case study is
to analyze the conditions which were present over Kansas and Oklahoma, and analyze the
evolution of convection to better understand why only one storm was able to produce
several tornadoes, including two significant tornadoes.
Synoptic Setup
At 1200 UTC a 500mb mid-tropospheric shortwave trough was located over northern
Arizona, southern Utah, and southern Nevada (Figure 4a). In response to this system,
pressure falls occurred during they day across portions of northwest Oklahoma, the Texas
panhandle, and southwest Kansas, causing a weak surface low to form in that area. A
warm front was slowing lifting north across northern Oklahoma and southern Kansas. A
cold front extended across northwestern Kansas northeastward into southeast Nebraska
and central Iowa. A dryline was already in the central Texas panhandle, while Amarillo
had a dewpoint of 54°F and Gage had a dewpoint of 66°F. Dewpoints were already in
the mid 60’s and lower 70’s across northern Oklahoma and southern Kansas (Figure 5).
A 700mb dry punch (Figure 4b) can be seen in New Mexico and entering the Texas
panhandle. Also, the westerly component is aiding in the eastward mixing of the dryline
as momentum from the 20 knot westerly winds are mixed down through the boundary
layer.
The 1200 UTC sounding from Lamont, Oklahoma (LMN) (Figure 6) in extreme north
central Oklahoma shows a deep layer of moisture from the surface up to 850 mb, with an
850 mb dewpoint temperature of 15°C. It also shows a capping inversion between 850
mb and 820 mb, and also a mid-level mixed layer near the 700mb level. A strong veering
wind profile is also noted in the lowest 0-6 km, indicating warm air advection. This
sounding has most of the classic characteristics of the Fawbush-Miller “loaded gun”
sounding (Schaefer, 1986). This capping inversion will keep convection from initiating
until later in the afternoon, when the shortwave reaches the area, and daytime heating
warms the surface temperatures to near the convective temperature.
As the 500mb shortwave trough approached the Kansas/Oklahoma area, the surface
warm front slowly lifted north to just north of the Kansas/Oklahoma border. By 2300
UTC the surface front was now a slow moving cold front. The dryline can clearly be
seen extending into western Oklahoma and southwest Kansas (Figure 7). At near this
time thunderstorms were initiating along the dryline in southern Kansas. A radar image
from ICT (Wichita) at 2333 UTC (Figure 8) shows two cells have already formed along
the dryline, with younger cells to the north forming along the cold front.
At 0000 UTC the 500 mb shortwave trough was located over the four corners region of
the southwestern U.S. The 500 mb winds had increased to 40 knots from the southwest
in Dodge City, Kansas, and the 700 mb winds had increased to 35 knots out of the southsouthwest. The 850 mb wind observation at Dodge City is not representative of the
supercell-environment since it is located on the frontal boundary at that time. At 300 mb
the wind speed was generally 45-50 knots across Kansas, which is generally a strong
upper-level wind speed, helping to keep rain away from the updrafts of thunderstorms
(Fawbush et al. 1951). Easterly winds at the surface in eastern Colorado ahead of the
cold front aided in orographic lift in the upslope flow, helping to develop the supercells
which occurred in that region.
Radar Analysis
At 2333 UTC (Figure 8a) two main cells had initiated along the dryline in south-central
Kansas. The southern-most cell was located in central Barber County, while the cell to
its northwest was located in northwest Barber County. Field observations indicate that
while the northern cell showed signs of organization and supercell structure, the left-split
or anticyclonic split of the southern cell helped to cut off the inflow to the northern cell,
which was eventually the demise of the northern cell.
By 0008 UTC (Figure 8b), three main cells were ongoing across south-central Kansas.
The southern most supercell storm was located in eastern Barber County, and was just
beginning to produce the first tornado of 13 in its life. The tornado was on the ground
between 0007 UTC and 0020 UTC (Picture 1 and 2). The northern most supercell was in
the process of storm-splitting, and produced a tornado near Pratt at 0001 UTC. Between
0034 UTC and 0056 UTC base reflectivity and field observations indicate that the old
low-level mesocyclone was in extreme eastern Barber County while a new, and rapidly
intensifying, mesocyclone was forming in western Harper County (figures 8c, 8d, and
8e). The two hooks on base reflectivity indicate the approximate locations of the old and
new mesocyclones. The new mesocyclone produced an F3 tornado which damaged
property in Attica. By 0100 UTC yet another low-level mesocyclone was forming to the
east of the old low-level mesocyclone. This mesocyclone produced at least two multiple
vortex tornadoes during its lifetime. The hook associated with the mesocyclone can be
seen on base reflectivity in Figure 8f. By 0110 UTC a large wedge shaped tornado was
observed near Harper, KS (Picture 3). The tornado was nearly stationary, drifting slowly
east-northeast. Tilt one of the storm relative velocity product shows an impressive
velocity couplet, with 50 knot winds going towards the radar and 50 knot winds going
away from the radar (Figure 8h).
By 0144 UTC (Figure 8g) base reflectivity shows an impress hook echo, a radar signature
which implies a low-level mesocyclone is present. Field observations indicate rain was
beginning to wrap around the back side of the low-level mesocyclone, obscuring the view
of the tornado from the northwest. This supercell, which indicated classic supercell
characteristics, was now evolving into a high precipitation (HP) supercell. The cold front
can be seen to the north where the convection is forming a fine line, beginning to slowly
catch up to the cyclic tornadic supercell.
Two more tornadoes were reported between 0208 UTC and 0214 UTC in Harper county.
More widespread convection was occurring west and northwest of the supercell along the
cold front, along with convective clusters forming off of the low-level 850 mb jet. By
0333 UTC, the cold front had caught up with the supercell and a line of thunderstorms
extended from Emporia to near Medicine Lodge, KS (Figure 8i). This was the beginning
of a mesoscale convective system when moved east into western Missouri. Meanwhile,
convective storms continued to form over south central Kansas between 0530 UTC and
1100 UTC.
Vertical Shear Profile Analysis
I already know tornadic supercells formed across southern Kansas by 0000 UTC on May
12, 2004. I have already determined, based on previous research, modeling and personal
observations that 20-30 knots of 0-6 km shear is needed to support supercell
thunderstorms. The 0000 UTC DDC sounding (Figure 9) was behind both the surface
and 850 mb cold front. Therefore the indicated wind fields for those pressure levels are
not representative of the supercell environment. However, I due believe the 700 mb
winds and above are fairly accurate in representing the storm environment. The sounding
reveals a 700 mb wind of 30 knots from the south-southwest, and a 500 mb wind of 35
knots from the southwest. However, 6 km may be better represented by the wind
observation near 400 mb, which is 40 knots. Assuming a surface wind from the southeast
(at 15 knots), there was approximately 40 knots of 0-6 km shear, which would certainly
support supercells. Also, the shear did increase over the next two hours while significant
tornadoes were occurring. Interestingly, the sounding calculations show 55 knots of 0-6
km shear, which is very strong deep layer shear. Although, it is based on a northeast
surface wind, which is probably why the shear magnitude is so high. I am not entirely
sure how this is calculated, but it at least clear that with the degree of veering over 0-6
km and the speed shear involved, the environment was supportive of supercells.
In order to analyze the low-level shear environment, I will refer to the Haviland, KS
profiler (Figure 10). Between 0000 UTC and 0100 UTC the main tornadic supercell was
nearly about 40 miles from the profiler. Between 2300 UTC and 0100 UTC the 850 mb
winds backed substantially from the southeast at 10 knots to the east-southeast at 15
knots, respectively. Although, this is probably due to the cold frontal passage, since the
“fine line” on radar can be seen two counties to the east of Haviland, sloping towards
Haviland. The 700 mb winds increased from 30 knots to 40 knots. The upper level wind
speeds generally remained between 40 and 45 knots, with no significant changes in
magnitude. Since it was the 0-3 km shear that increased by 15 knots, this may be one
explanation for the increase in, not only observed low-level mesocyclone intensity, but
also the increase in tornado intensity, with time. More specifically, the 700 mb winds
increased 10 knots between the time of the first weak tornado at 0007 UTC and the time
of the F4 tornado 0120 UTC. 700 mb wind magnitude could play a role in significant
tornado potential.
Instability
There is some balance between CAPE and vertical wind shear (0-6km) that is necessary
to support most supercell thunderstorms. In Dodge City, Kansas at 0000 UTC May 13
(Figure 9), there was 1931 J/kg of surface based CAPE. But that was behind the cold
front, so the amount of CAPE further southeast must have been substantially greater.
Based upon the RUC analysis at 0000 UTC, there was somewhere between 3000 and
4000 J/kg of mixed layer CAPE. While that is not the same as surface based CAPE, it
still provides a clue that there was likely more than 3000 J/kg of CAPE in the supercell
environment. This may have been an important factor in causing an intense low-level
mesocyclone, which produced an F4 tornado near Harper, Kansas.
Conclusion
On May 12, 2004 13 tornadoes were produced from the one cyclic tornadic supercell.
The supercell formed along the dryline in south central Kansas. Other supercells did
form; however, the southern most storm became the cyclone tornadic supercell. This was
mainly due to the abundant moisture supply from the south. The 700 mb winds increased
by 10 knots between 2300 UTC and 0100 UTC, which may have played a role in the
supercell’s ability to produce a violent tornado. It may be significant when the winds
over the lowest 3 km increase in magnitude, even if it’s only 10 or 15 knots. It is still
clear that low-level shear and helicity are crucial factors for the development of low-level
mesocyclones, and hence tornadoes, but the optimum values are still not entirely known
(Davies and Johns, 1993).
Figure 3: Preliminary storm reports for May 12, 2004 from the Storm Prediction Center.
Figure 4a: 500 mb analysis at 1200 UTC on May 12, 2004.
Figure 4b: 700 mb analysis at 1200 UTC on May 12, 2004.
Figure 5: National surface analysis at 1200UTC on May 12, 2004.
Figure 6: 1200 UTC morning sounding on May 12, 2004 from Lamont, OK.
Figure 7: 2300 UTC surface analysis on May 12, 2004. The dryline and cold front are analyzed in brown
and blue, respectively.
Picture 1: Taken at 0010 UTC, a tornado in eastern Barber county on May 12, 2004 produced by the
southern supercell in Kansas (photo by Simon Brewer).
Picture 2: About seven minutes later in almost the same location as Picture 1 (photo by Simon Brewer).
Figure 8a: ICT base reflectivity at 2333 UTC. The early stage of the cyclic tornadic storm can be seen in
extreme southern Kansas.
Figure 8b: Wichita (ICT) base reflectivity at 0008 UTC on May 13, 2004 .
Figure 8c: ICT base reflectivity at 0034 UTC on May 13, 2004. The southern supercell is the cyclic
tornadic one. The cold front is evident by the “line” of convection to the northeast.
Figure 8d: ICT base reflectivity at 0051 UTC on May 13, 2004.
Figure 8e: ICT base reflectivity at 0056 UTC on May 13, 2004. A double hook structure is evident on the
southern supercell, indicating a new and old mesocyclone are both present.
Figure 8f: ICT base reflectivity at 0118 UTC on May 13, 2004.
Figure 8g: ICT base reflectivity at 0144 UTC on May 13, 2004. A large hook echo is evident on the
southern supercell.
Figure 8h: ICT storm relative velocity tilt 1 at 0125 UTC on May 13, 2004. A strong velocity couplet of 40 knots and +50 knots can be seen in eastern Harper county (circled in white).
Picture 3: Video grab of a tornado observed at 0115 UTC just southwest of Harper, KS looking south
(video capture by Jim Bishop).
Figure 8i: Regional radar composite at 0333 UTC on May 13, 2004.
Figure 9: Dodge City (DDC) 0000 UTC sounding from May 13, 2004. Notice the low-level winds are
from the northeast, indicating the sounding was launched behind the cold front.
Figure 10: The Haviland, Kansas profiler between 0900 UTC May 12, 2004 and 0600 UTC May 13, 2004.
Haviland is located in Kiowa county, about 40 miles northwest of the cyclic tornadic supercell’s path.
Case II: May 22, 2004 Central Plains Tornado Outbreak
On May 22, 2004 over 80 tornadoes were reported across southeastern Nebraska and
central Iowa (Figure 11a). Tornadoes were also reported in eastern Colorado,
southeastern Wyoming, Wisconsin, Indiana and Michigan. The most significant
supercell formed in Nuckolls County in south central Nebraska. As it moved northeast, it
produced a long-track tornado which grew to an unprecedented two and one half miles
wide at its widest point, causing F4 damage just east of Wilber, Nebraska. This is now
the largest tornado on record. There were also numerous supercells in the region
producing tornadoes all evening. This event was truly a historic tornado outbreak. The
purpose of this case study is to understand the conditions which favored such a large
number of tornadoes, and especially to understand what caused such a large tornado.
Synoptic Setup
At 1200 UTC May 22, 2004, a 500 mb longwave trough (Figure 11) extended across the
western half of the contiguous U.S. An embedded mid-tropospheric shortwave trough
was located over the Great Basin region. The coldest temperatures were located over
Nevada with a core of -20°C. An imbedded speed maximum had just entered New
Mexico, with a 50 knot wind observation at the leading edge of the jet, and a 55 knot
wind observation in southern Arizona. With the approach of the shortwave trough, and
the imbedded speed maximum coming from the desert southwest, pressure falls were
occurring over the central high plains of Colorado and northern Kansas region. At 300
mb (Figure 12), a jet streak with a core of 105 knots was situated from southwestern
Arizona curving northeast into northeastern New Mexico and southeastern Colorado.
This jet streak ejecting into the central plains, coupled with the digging 500 mb jet
maximum, would aid in considerable rising motion over southern Nebraska later in the
afternoon.
The 1200 UTC 700 mb analysis (Figure 13) shows a temperature ridge located from
Mexico northward through Texas, Oklahoma, and extending into southern Nebraska.
The 12°C isotherm stretched across northern Kansas, with the 9°C isotherm in extreme
southern Nebraska.
The 1200 UTC 850mb (Figure 14) analysis showed a very moist environment with a
dewpoint of 14°C located in Kansas City, and 12°C at Dodge City. There was no
significant warm air advection evident on this analysis. A frontal boundary was oriented
east-west across southern Nebraska.
The 1200 UTC surface analysis revealed a 997 mb surface low over northern Kansas
(Figure 15), with an associated stationary from running from eastern Colorado eastnortheast through central Kansas, then extending northeast into Minnesota, and east to the
Lake Michigan. Temperatures ahead of the front were in the middle 60’s and lower 70’s,
with 50’s behind the front. Dewpoints ahead of the front were mainly in the lower 60’s,
while they were slightly lower across southern Nebraska, which was in the mid and upper
50’s. Southerly winds were already advecting moisture into the southern Nebraska
region (all surface temperatures are in °F).
The 1200 UTC Omaha (OAX) sounding indicates that a cold front had just passed
through, with a backing wind profile from the surface to 800 mb. The 1200 UTC Topeka
(TOP) sounding reveals deep boundary layer moisture already in place, with a weak
capping inversion located at 850 mb (both in Figure 16). The Dodge City (DDC)
sounding (also in Figure 16) indicates a stronger capping inversion at 850 mb, which will
be advecting into the southern Nebraska region by the afternoon hours.
By 1800 UTC, the frontal boundary was lifting north now as a warm front, which now
extended from southwest Nebraska east-northeast into central Nebraska, and then
continuing east into central Iowa as a stationary front (Figure 17). Easterly flow (upslope
regime) in southwestern Nebraska and northeastern Colorado was aiding in considerable
rising motion. This can be clearly seen from the 1800 UTC LBF sounding. Note the
upslope flow from the surface to 700 mb (Figure 18b). This, along with the approaching
shortwave trough from the Great Basin, would make this area to be the first area to have
thunderstorm initiation. Due to the imminent possibility of a tornado event in the central
plains, several NWS offices across the region also launched 1800 UTC special soundings.
The 1800 UTC DDC sounding (Figure 18a) shows that the nose of the 500 mb jet had
made its way into southwest Kansas, with a 50 knot wind observation at 500 mb. Also, a
65 knot wind was observed at 300 mb, indicating that the 300 mb jet streak was in the
area as well. Soon this will help to enhance lift over southern Nebraska. Dry adiabatic
lapse rates are present in the middle levels at DDC. It should be noted that DDC was
located behind the dryline at 1800 UTC. The 1800 UTC OAX sounding indicates a
stronger capping inversion, with a convective temperature of 91°F. This sounding shows
it is obvious no thunderstorms were likely to initiate along the warm front for another 2-3
hours. However, further west where the upper level support was starting to move into the
region, strong upslope flow aided in rising motion, and surface temperatures were getting
closer to the convective temperature, thunderstorm initiation was likely soon.
At 1900 UTC, the RUC’s analysis of surface based CAPE showed a large area of 3000
J/kg covering all of southern Nebraska, with greater values further south (Figure 19). It
still had between -25 and -75 CIN values over the area, so the cap was still holding
strong. By 2016 UTC, the dryline had bulged east in central Kansas, with dewpoints in
the 30’s°F behind the dryline and dewpoints in the upper 60’s°F ahead of the dryline.
The warm front was still situated basically east-west across southern Nebraska, with a
surface low pressure center in west-central Kansas near Hill City (Figure 20). Rich low
level moisture was in place with dewpoints in the middle and upper 60’s°F across
northern Kansas and southern Nebraska. The high plains of western Nebraska and
northeastern Colorado had much lower dewpoints. But moisture advection from the east
was occurring, and the area would soon be supportive of tornadoes. The rich 850 mb
moisture can be seen in Figure 21. Notice the pooling of 10°C - 14°C 850 mb dewpoint
temperatures that is occurring across southern Nebraska along the warm frontal
boundary. Also notice how this rich moisture has spread into western Nebraska, where
supercell thunderstorm initiation is becoming increasingly likely.
The 500 mb shortwave had made its way to Colorado by 0000 UTC May 23, 2004
(Figure 22), with the coldest temperatures aloft located over Colorado and Utah. The 50
knot speed maximum had made it to Dodge City Kansas. This would place the left exit
region of the jet streak over western Kansas, which may have helped to initiate the
supercells over that region. The shortwave can also be seen at 700 mb over eastern
Colorado and New Mexico, with the warm front even evident at this level in southern
Nebraska (Figure 23). The 850 mb analysis at 0000 UTC reveals impressive dewpoint
temperatures of 16°C across southern Nebraska and northern Kansas (Figure 24). Omaha
recorded a 16°C 850mb dewpoint temperature, and so did Kansas City. The warm front
is also evident somewhere between northern Kansas and southern Nebraska. Based on
the RUC’s 1900 UTC analysis, the warm front was probably still over southern
Nebraska. The 0000 UTC Omaha (OAX) sounding (Figure 26a) reveals the station was
just north of the warm front, with northeasterly winds over the lowest 1 km. It was a very
unstable environment with 3668 J/kg of observed CAPE. The cap had completely been
eroded by a combination of isentropic lift along the warm front, and upper level support
from the 500 mb shortwave. However, the environment was still capped further south in
Topeka (TOP) where the 800 mb temperatures were much warmer (Figure 26b). Nearly
3500 J/kg of CAPE was also observed at this location. It should be mentioned that the
0000 UTC TOP sounding is nearly the perfect example of the classic Fawbush-Miller
“loaded gun” sounding (Schaefer, 1986), which is typically observed several hours before
a tornado outbreak.
Radar Analysis
By 2100 UTC on May 22, 2004, a combination of easterly upslope flow along the high
plains of western Nebraska, and isentropic assent along the warm front caused supercells
to initiate across northwestern Kansas and western Nebraska (Figure 27). Within an hour
later, several supercells were in existence across western Nebraska, and two new
supercells had just initiated off the warm front across eastern Nebraska. One tornado had
already been reported in McPherson County northwest of North Platte, Nebraska. Within
minutes other supercells further southeast and south of North Platte began producing
tornadoes (Figure 28). The supercells in eastern Nebraska had strong mid-level rotation,
but weak low-level rotation based on field observations. Subsequently, they were not yet
producing tornadoes.
It was now 2332 UTC and significant supercell development was finally occurring across
the eastern half of Nebraska, along with west-central Iowa (Figure 29). Numerous
supercells had formed along the warm frontal boundary in Iowa, and the supercells in
eastern Nebraska were intensifying. The cold front was surging southeast out of the high
plains of western Nebraska, causing a squall line or thunderstorm complex in western
Nebraska. Strong isentropic lift along the warm front was causing scattered showers and
storms across central Nebraska. But the most significant development was a new
supercell which had recently formed along the intersection of the dryline and an outflow
boundary in Nuckolls County in south central Nebraska. This particular supercell was
completely isolated, and was in the best low-level shear environment.
At 2336 UTC Figure 30 reveals the location of the now retreating dryline and an outflow
boundary to its northwest. This supercell had already tracked to the northeast several
miles by this time and was now located in northwest Thayer county. But it is apparent
that this storm initiated along or just ahead of the dryline where it bulges to the east, and
where the dryline intersects the west-southwest/east-northeast oriented outflow boundary.
The surface map from 0000 UTC May 23 (Figure 25) indicates an east wind at 20 knots
located well to the west of the outflow boundary. Also, the cold front is noted in
southwest Nebraska with northerly winds between 25 and 30 knots. Therefore, the new
supercell which initiated in western Nuckolls County did in fact form near the
intersection of the dryline bulge and an outflow boundary which, at 2336 UTC was
located near the intersection of Webster/Nuckolls counties and the Nebraska/Kansas state
border. Note the hook echo on the supercell in Franklin County, located three counties
west of the new supercell in Thayer County. A tornado was produced from this supercell
at the time of this radar image in Franklin County, 1 mile west of Campbell, Nebraska.
Ten minutes later the same supercell produced another tornado one county to the east in
Webster County near Bladen, Nebraska.
At 2352 UTC (Figure 31) note the dryline has retreated to the west since the last radar
image (16 minutes). Also, the outflow boundary has moved further south and east. A
tornado was reported at the time of this radar image 3 miles northwest of Hebron,
Nebraska in northern Thayer County. This tornado was the first tornado of at least two
tornadoes that caused about 60 miles of damage. Note the supercell has split into a left
and right split cell. The left (anti-cyclonic) supercell is located to the northwest of the
tornadic, cyclonic supercell in northern Thayer County. Also note that the other tornadic
supercell to the west, which is now in Webster County, no longer has a clearly visible
hook echo, although a wet hook can still be seen. By 0003 UTC, May 23, a hook echo is
clearly visible associated with the low-level mesocyclone with the Thayer county
supercell (Figure 32). A tornado had been on the ground for eleven minutes and was now
reported near Bruning Nebraska in Thayer County. This is the last radar image which
shows indications of this particular supercell having classic supercellular characteristics,
such as a hook echo. The supercell then began evolving into a more high precipitation
supercell.
By 0019 UTC (Figure 33a), notice how the supercell appears completely HP in nature.
Also, the dryline has continued to retreat to the west, approaching the outflow boundary
to its northwest. Notice that tilt 1 of the SRV indicates a couplet of strong rotation just
east of the “wet” hook (Figure 33b). However, the base reflectivity image reveals that
this area of rotation is located in the same area where 45 DBZ is indicated.
Consequently, this low-level rotation is wrapped in rain, yet another indication that this
supercell has evolved into a high precipitation supercell. The tornado was still on the
ground at this time. Twenty minutes later (0040 UTC, Figure 34), the storm had gone
through storm splitting again, and the anti-cyclonic split supercell is noted to the
northwest of the cyclonic split supercell.
Around 0040 UTC, May 23, the right split supercell (cyclonic) produced what became
the largest tornado in history (Figure 34). Based on the NWSFO Omaha, Nebraska
damage survey the tornado initially touched down in northwest Jefferson County (Figure
35). Notice the appendage on the southwestern edge of the reflectivity, where there is
still a “wet” hook echo, still an indication that rain is completely wrapping around the
mesocyclone. Figure 36a shows the close-up of the base reflectivity at 0107 UTC (OAX,
Omaha). A donut shaped reflectivity pattern is seen just south of Wilber, Nebraska,
indicating an intense mesocyclone completely wrapping precipitation around it. In
Figure 36b (SRV tilt1), a very intense mesocyclone is evident. A large couplet of
approximately 40 knots inbound and 40 knots outbound is an indication of a very large
mesocyclone. The smaller couplet can be seen just south-southwest of Wilber, with
approximately 60 knots outbound and 60 knots inbound. Photo 3 is a video grab taken at
about the same time as the radar image, looking south-southwest from a location
approximately ½ mile east of downtown Wilber. The low-level mesocyclone was rapidly
increasing in size at this time. The tornado was only visible from north-northeast and due
north of the tornado due to rain wrapping around the mesocyclone, obscuring the view
from all other directions. This fits very well with the reflectivity image at this time, as it
can be see there is an indication of an inflow notch southeast of Wilber. However, this
image is not representing the lowest levels of the atmosphere, so I am confident this
image is showing “some” of the precipitation that is being suspended in the air in the
BWER (Bounded Weak Echo Region), or the vault, of this high precipitation supercell.
Otherwise, observation of the tornado would have been impossible from that location at
that time.
Between 0107 UTC and 0124 UTC, the tornado grew to its widest size, an unprecedented
2.5 miles wide. The tornado was observed to be a very large wedge at 0112 UTC (photo
4) and still growing in size. By the 0124 UTC the tornado was east-northeast of Wilber
by about 1 mile, however a short lived satellite tornado was observed over downtown
Wilber at approximately 0120 UTC, damaging a few buildings. A wet hook can be seen
just northeast of Wilber, associated with the very intense mesocyclone (Figure 37a).
Figure 37b indicates a very intense and impressive SRV couplet, with 80 knots outbound
and 80knots inbound. The size of the SRV couplet is incredible. This goes along very
well with what Weisman and Klemp (1982) said, that the existence of a strong low-level
mesocyclone is closely related observationally to the development of tornadoes. The
tornado continued moving northeast and caused F4 damage in the town of Hallam,
Nebraska, located about ten miles northeast of Wilber. After Hallam the tornado finally
shrunk in size and began to weaken. The tornado’s path length was 52 miles long, was
2.5 miles wide at its widest point, and created F4 damage in Hallam, Nebraska.
By 0330 UTC, the convective activity had congealed into a squall line extending from
north central Kansas northeast into central and eastern Iowa. Some of the most intense
cells were supercells imbedded in the line, and did produce tornadoes. Isentropic lift
continued north of the front in central and northern Nebraska and all of northern Iowa,
creating moderate rain in those areas.
Shear Analysis
The first supercells of the day in northeast Colorado and western and southwestern
Nebraska were ongoing by 2100 UTC. Based completely off of the 1800 UTC LBF
sounding, 45 knots of 0-6km shear was present, which is definitely supportive of
supercell thunderstorms. Only 10 knots of 0-3km shear existed, with no direction shear
present. So the winds must have increased in speed and veered quite a bit in the middle
levels for tornadoes to have formed.
Both the OAX and the TOP soundings show there was between 40 and 60 knots of 0-6
km shear present at 0000 UTC, May 23, 2004 (figure 26 and 27). But a closer analysis of
the lowest 0-3 km is necessary to establish the shear environment that may have affected
the tornado’s size and intensity between 0000 UTC and 0130 UTC. The Fairbury,
Nebraska wind profiler, located in southeastern Nebraska about 20 miles southwest of the
Wilber/Hallam tornadic supercell, reveals some interesting things about the wind profile
(Figure 38). Between 0000 UTC and 0100 UTC, the 0-3 km shear increases
dramatically. The 850 mb winds increase from 30 knots at 0000 UTC to 45 knots at 0100
UTC. There is a large increase in wind magnitude in the 700 mb winds as well, from 30
knots at 0000 UTC to 50 knots at 0100 UTC. Interestingly, the 500 mb winds also
increased in magnitude, from approximately 55 knots at 0000 UTC to approximately 70
knots at 0100 UTC. That resulted in incredibly strong 0-6 km shear, given that the winds
below 1 km were backing to the south-southeast. With a 50 knot wind at 700 mb or
approximately 3 km, and a surface wind near 20 knots, that’s approximately 30 knots of
0-3 km shear, which is very impressive. If you compare this to the 10-15 knots of 0-3km
shear that was present at 0000 UTC, the shear doubled in magnitude, possibly owing to a
much stronger tornado. In addition the 500 mb winds increased by 15 knots between
0000 UTC and 0100 UTC. This enhanced the overall shear for the supercell itself,
possibly having a positive effect on the low-level mesocyclone as well. Overall, the deep
layer shear was very strong. That, coupled with ~3500 J/kg of CAPE resulted in very
intense supercell thunderstorms.
The OAX 0000 UTC sounding does indicate that the 3 km storm relative helicity was 225
m^2/s^2. That, of course, was calculated from wind observations before the 0-3 km
shear increased by approximately 15 knots. The 3 km SRH likely increased dramatically
between 0000 UTC and 0100 UTC, which is probably one reason why a violent tornado
occurred. This same idea was mentioned by Moller et al. (1994) who stated that the 3 km
SRH is crucial for determining mesocyclone potential. This could be interpreted such
that stronger 3km SRH values indicate a higher likelihood of a mesocyclone, possible
even strong mesocyclones. Since strong low-level mesocyclones are related to strong
tornadoes, it is possible that high 3 km SRH values are an indication of the potential for
strong tornadoes.
By 0200 UTC, May 23, the 850 mb winds had veered to the southwest, which was 45
degrees from what their direction the previous hour. This is likely the reason why the
convective mode was linear at this time. However, since the 0-1 km shear values were
still very high due to the 50 knot 850 mb winds, imbedded supercells and mesocyclones
were present within the squall line. Another reason for the linear storm mode is the
forcing. The cold front was rapidly moving to the southeast at 0000 UTC from southwest
Nebraska. It likely spawned the squall line, due to very strong linear forcing.
Conclusion
On May 22, 2004 approximately 88 tornadoes occurred across Colorado, Nebraska,
Kansas, and Iowa. One of which was the widest tornado ever recorded, reaching an
unprecedented 2.5 miles wide, and causing F4 damage in the small town of Hallam,
Nebraska. It has been shown that orographic lift over the high plains of southwestern
Nebraska and northeastern Colorado, coupled with warm surface temperatures and upper
level support, caused tornadic supercells to initiate in that region. It has also been shown
that tornadic supercells formed along the warm front across eastern Nebraska and central
Iowa. Supercells formed along the dryline in northern Kansas. But most significantly,
the most impressive tornadic supercell formed along the intersection of an outflow
boundary and a dryline bulge in southern Nebraska.
Deep low-level moisture, high CAPE values around the 3500 J/kg range, and strong
shear, in both the 0-6km layer and the 0-3km layer, helped to support tornadic supercells
in central and southeastern Nebraska. The strong increase in the 3 km shear likely played
an important role in the development of the large, significant tornado which went through
Wilber and Hallam, Nebraska, causing F4 damage in Hallam. Why this particular shear
environment was able to produce a 2.5 mile wide tornado is unknown at this time.
Figure 11a: Storm reports from May 22, 2004 from the Storm Prediction Center.
Figure 11: 500 mb analysis at 1200 UTC on May 22, 2004.
Figure 12: 300 mb analysis at 1200 UTC from May 22, 2004.
Figure 13: 700 mb analysis at 1200 UTC on May 22, 2004.
Figure 14: 850 mb analysis at 1200 UTC on May 22, 2004.
Figure 15: Surface analysis at 1200 UTC on May 22, 2004.
Figure 16: The above are all sounding launched at 1200 UTC May 22, 2004. Top left: Omaha,
Nebraska. Top right: Topeka, Kansas. Bottom left: Dodge City, Kansas.
Figure 17: Surface analysis at 1800 UTC on May 22, 2004.
Figure 18a
Figure 18b
Figure 18: Figure 18a is the Dodge City sounding launched at 1800 UTC on May 22, 2004. Figure 18b is
the North Platte, Nebraska sounding launched at 1800 UTC on May 22, 2004.
Figure 19: RUC surface based CAPE and CIN (J/kg) analysis at 1900 UTC from May 22, 2004.
Figure 20: Surface plot at 2016 UTC from May 22, 2004. Notice how the frontal boundary has stalled
over southern central Nebraska.
Figure 21: 1900 UTC RUC 850 mb analysis from May 22, 2004. Black-height, brown-temp, greendewpoint.
Figure 22: 500 mb analysis at 0000 UTC from May 23, 2004
Figure 23: 0000 UTC 700 mb analysis from May 23, 2004.
Figure 24: 0000 UTC 850 mb analysis from May 23, 2004.
Figure 25: Surface plot at 0000 UTC from May 23, 2004. Notice the cold front surging south in
southwestern Nebraska, which is being aided by convection.
Figure 26a: Omaha, Nebraska 0000 UTC sounding from May 23, 2004. The northeasterly winds in the
low-levels indicate this sounding was launched north of the frontal boundary.
Figure 26b: Topeka, Kansas sounding at 0000 UTC from May 23, 2004. Notice the capping inversion
still present just above 850 mb.
Figure 27: Regional radar composite of base reflectivity from May 22, 2004 at 2100 UTC.
Figure 28: Regional radar composite of base reflectivity from May 22, 2004 at 2202 UTC.
Figure 29: Regional radar composite of base reflectivity from May 22, 2004 at 2332 UTC.
Figure 30: Hastings, Nebraska (UEX) base reflectivity from May 22, 2004 at 2336. The dryline can be
seen as a fine line on radar sloping southwest in Kansas, and an outflow boundary can also be seen oriented
northeast/southwest across the Kansas/Nebraska border. Two tornadic supercell can be seen, both in
southern Nebraska. The Franklin county supercell has a hook echo. Another tornadic supercell is located
in Thayer county, while another supercell has been ongoing across east-central Nebraska to the northeast.
Figure 31: Hastings, Nebraska base reflectivity from May 22, 2004 at 2352 UTC. The Thayer county
supercell (as shown in Figure 30) has split into a right (cyclonic) and left (anti-cyclonic) split supercell,
which can both be seen in southeast Nebraska. Also, the dryline is retreating to the west.
Figure 32: Hastings, Nebraska base reflectivity from May 23, 2004 at 0003 UTC. The right split supercell
(which produces the tornadoes, including the 2.5 mile wide tornado) now has a hook echo. Also, the
dryline (fine line in the image) continues retreating to the west.
Figure 33a: Hastings, Nebraska base reflectivity from May 23, 2004 at 0019 UTC.
Figure 33b: Hastings, Nebraska storm relative velocity tilt 1 from May 23, 2004 at 0019 UTC. The
supercell now has a strong velocity couplet which as been circled in white.
Figure 34: Hastings, Nebraska base reflectivity from May 23, 2004 at 0040 UTC.
Figure 35: Path of the tornado which passed through the south side of Wilber, Nebraska and destroyed the
town of Hallam, Nebraska on May 22, 2004. Just east of Wilber the tornado grew to an unprecedented 2.5
miles wide, which makes this tornado the widest tornado on record.
Figure 36a: Omaha, Nebraska (OAX) base reflectivity from May 23, 2004 at 0107 UTC. Notice the
“donut” shaped reflectivity pattern, an indication of an intense mesocyclone associated with a high
precipitation supercell.
Figure 36b: Omaha, Nebraska storm relative velocity tilt 1 from May 23, 2004 at 0107 UTC. Notice the
strong velocity couplet. The couplet is located due south of Wilber. Photo three (a video capture) was
taken at about the same time as this radar image.
Photo 3: Taken at ~0110 UTC on May 23, 2004 from less than ½ mile east of downtown Wilber,
Nebraska looking south (video capture by Jim Bishop).
Photo 4: Video capture taken at ~0115 UTC, same location as Photo 3, looking south. The tornado is now
more than one mile wide, possibly nearly two miles wide at this point, and continues to grow (video
capture by Jim Bishop).
Figure 37a: Omaha, Nebraska base reflectivity from May 23, 2004 at 0124 UTC. Notice the supercell’s
wet hook echo, a radar characteristic of a high precipitation supercell. Wilber is located in the hook.
Figure 37b: Hastings, Nebraska storm relative velocity tilt 1 from May 23, 2004 at 0133Z. The intense
velocity couplet is located directly over Hallam.
Figure 38: Fairbury, Nebraska profiler, between 0900 UTC May 22 and 0800 UTC May 23, 2004.
Fairbury is located in extreme south central Nebraska.
Case III: May 29, 2004 Southern and Central Plains Tornado Outbreak
On Saturday, May 29, 2004 a classic, significant tornado outbreak unfolded across
portions of Oklahoma, Kansas, Nebraska and northern Missouri. Approximately 90
tornadoes were reported across those states (Figure 39a), but these tornadoes were
produced by only a few cyclic tornadic supercells. There are two specific reasons for
doing a study on this event. The first is this was a classic setup for a tornado outbreak,
thus it was forecasted well in advance. The second reason is there were two damaging
supercells; one in Oklahoma and the other in southern Kansas. However, the southern
Kansas supercell was mainly a classic, which produced over a dozen photogenic
tornadoes (some long-track), while the Oklahoma supercell produced several short-track
tornadoes, and was HP in nature during the majority of its life. In this case study, I hope
to reveal clues as to why the southern Kansas supercell produced photogenic tornadoes,
and why the Oklahoma supercell became classic so much later in its life.
Synoptic Setup
On the morning of Saturday, May 29, 2004 at 1200 UTC, a longwave trough was situated
over the Contiguous U.S. at 500 mb. A negatively tilted shortwave trough was located
over the western half of the U.S. (Figure 39), with an associated trough axis stretching
from the Great Basin to New Mexico. The coldest 500 mb temperatures were in the
lower and middle -20’s ° F over Washington and Idaho. The -10 °C line stretched from
central New Mexico northeast through Dodge City and Kansas City. Strong flow was
already present across much of the southern plains, with 40 knot flow already in
Oklahoma and west Texas, while stronger 50 knot winds were still coming around the
southern periphery of the shortwave trough axis near the Mexican border. At 300 mb
(Figure 40), a 70 knot jet streak was located from west Texas northeast into southwest
Kansas and northwest Oklahoma. A 700 mb trough axis was oriented north-south across
eastern Montana, Wyoming, and Colorado (Figure 41). Despite warm temperatures
between 11 °C in Amarillo and 14 °C in LBF, cold air advection was already occurring
over the region. At 850 mb (Figure 42), a low-level jet was present across all of the
central and southern plains states, with abundant moisture. Wind magnitudes between 40
and 50 knots, mainly out of the south, were present across Texas, Oklahoma, Kansas and
Nebraska. Oklahoma and Kansas were both covered with 850 mb dewpoints above 12
°C. The south-southwesterly wind observation in Amarillo suggests that the dryline may
already be mixing to the east across the Texas panhandle.
The 1200 UTC Midland sounding (Figure 43a) has a classic cap and an associated
elevated mixed layer. This air will advect into Oklahoma and Kansas later in the day.
Meanwhile, the AMA sounding (Figure 43b) indicates a dry mid and upper atmosphere,
which bodes well for surface heating throughout the day across Kansas. Figure 44a and
44b are the Norman, Oklahoma and Lamont, Oklahoma soundings, respectively. Both
soundings show a very deep layer of moisture, along with an elevated mixed layer near
the 700 mb level. The air is saturated at 300 mb, which is an indication that upper level
clouds are in the area. However, soundings further upstream in Amarillo and Midland
reveal high dewpoint depressions in the upper levels, indicating a lack of upper level
clouds in those regions.
The Dodge City (DDC) sounding also revealed a deep layer of cirrus clouds at 1200 UTC
(Figure 45a), with an elevated mixed layer present around 800 mb. The deep low-level
moisture was already present as far north as Omaha, Nebraska at 1200 UTC (Figure 45b).
The surface map at 1311 UTC (Figure 46) showed a cold front which was located in
northeast Colorado and western Nebraska, slowly moving southeast. A stationary front
extended from eastern South Dakota southeastward into the Tennessee Valley, while a
pseudo warm front was oriented east-west across southern Oklahoma and into central
Arkansas. A dryline extended north-south from the central Oklahoma panhandle southsouthwest into west Texas. Dewpoints across Oklahoma were in the middle 60’s °F,
while Kansas and Nebraska had dewpoints in the middle 50’s °F.
By 1800 UTC (Figure 47) warm air advection had already increased the dewpoints across
most of Oklahoma into the lower 70’s (in °F). In western Oklahoma, just ahead of the
dryline, dewpoints were in the lower 60’s. The dewpoints in all of central Kansas had
already increased to the middle 60’s, with some lower 70’s already being reported in
southern Kansas. Meanwhile, middle 60’s were present in Nebraska. Surface
temperatures throughout most of the plains were already in the middle and upper 80’s,
while the temperatures in the Texas panhandle behind the dryline were already in the
lower 90’s. Note the relatively weak surface winds behind the dryline, indicating the
dryline was mot very strong at that time.
By 2100 UTC (Figure 48), Liberal Kansas was reporting a 30 knot wind out of the
southwest, with a 32 °F dewpoint temperature. The winds in the central Texas panhandle
had increased as well, now reporting southwesterly surface winds at 20 knots. These are
indications that the dryline had been, and was now mixing east. Later, radar imagery will
show that storms had already fired along the dryline and were ongoing by this time across
much of western Oklahoma.
At 2200 UTC (Figure 49), the surface wind in Gage, Oklahoma had backed substantially
to the southeast at 30 knots. The surface winds have backed substantially over the last
hour in northern Kansas and southern Nebraska in response to the surface low now over
north central Kansas.
By 0000 UTC the dryline has made it east of Gage, Oklahoma (Figure 50). It has bulged
to the southeast of the surface low in north central Kansas, and the surface winds have
backed to the east-southeast in southeastern Nebraska. Overall, the surface winds across
Oklahoma, Kansas and southern Nebraska have backed some. Surface temperatures
across northwestern Kansas and northeast Colorado have fallen into the 50’s °F behind
the now surging cold front. The surface low has deepened to 986 mb over north central
Kansas. The RUC’s 0000 UTC mixed layer CAPE analysis (Figure 51) has 3000 J/kg of
CAPE over most of the areas, with a smaller area of 4000 J/kg analyzed over south
central Kansas.
Still on the 0000 UTC surface map (Figure 50) it appears like there is a faint bulge in the
dryline in southern Kansas. The surface winds across the northern Texas Panhandle to
Liberal Kansas are a little stronger and more veered than in the southern panhandle and
south. If this is in fact the case, then there could have been drier air just above the
surface, say near 925 mb, being entrained into the Harper county storm. This would
cause the evaporation of rain near the updraft, and a more classic to low precipitation
supercell structure. It would certainly aid in a more visually impressive storm, capable of
producing photogenic tornadoes if the tornadoes existed, which they did.
The negatively tilted 500 mb shortwave trough had ejected into the southern plains by
000 UTC, owing to strong 45 knot flow out of the southwest across Oklahoma and
Kansas (Figure 51). The coldest temperatures had shifted to the east, but were still
mainly over the Great Basin. At 300 mb, a strong diffluent region is evident across
northern Kansas, and probably did cause upward vertical motions which would be
enough to enhance updrafts, or help break a capping inversion. Also, the left entrance
region to a weak (80 knot core) jet streak was located over northern Oklahoma and
southern Kansas. Due to the weak accelerations it would produce aloft, I would not
consider this a significant enough area of convergence aloft to really affect the overall
outcome of convection. Probably the most significant feature is the magnitude of the 300
mb winds across southern Kansas versus central Oklahoma. Central Oklahoma is right
under the 80 knot jet streak, while southern Kansas is somewhere between the 80 knot
and 65 knot observations. Field observations from several different persons indicate that
the supercell which tracked across Harper County, Kansas was between being a classic
and a low precipitation supercell throughout most of its life. Conversely, personal
observations from the supercell which tracked across the entire state of Oklahoma
indicate that it was a high precipitation supercell throughout most of its life. Interestingly
enough, however, it appears the stronger 300 mb winds were located right over the
Oklahoma supercell, while the southern Kansas supercell had weaker 300 mb flow.
Brooks and Wilhelmson (1993) found that one of the effects of strong winds aloft is to
blow rain away from the updraft, which could play a role in storm type. Although, 300
mb winds are only one factor in several that determine storm type.
The 0000 UTC 700mb flow (Figure 54) is strong throughout the southern and central
plains, being 40 knots and out of the southwest. Another interesting feature is the
warmer 700 mb temperatures at OUN, AMA, and DDC as compared with the 1200 UTC
temperatures. Despite cold air advection at 1200 UTC and the apparent cooling in the
midlevels due to the moderate vorticity advection at 500 mb, down sloping and
compressional heating was stronger. The 850mb winds are also strong with 35 knots out
of the south-southwest at OUN (Figure 55). Kansas City has a 45 knot 850 mb wind
straight from the south. Dodge City’s observation is out of the south-southwest, but is
behind the dryline, and is not representative of the thunderstorm environment. The
Haviland wind profile (Figure 56) shows a 50 knot wind slightly west of due south at 850
mb. Given the strong moisture gradient between the dryline and the moist boundary
layer environment in the warm sector, this “slight” veering of the 850mb wind in south
central Kansas may have been ideal to advect the perfect amount of dry air into the storm,
keeping it a classic/low precipitation supercell throughout its entire life.
The 0000 UTC OUN sounding reveals a few interesting features that were present in an
area near the Oklahoma tornadic supercell (Figure 57). A 800 mb capping inversion was
still present at the time of the sounding. The available CAPE was 2500 J/kg, as apposed
to the RUC’s 3000-4000 J/kg analysis at this same time (Figure 51). Interestingly, the
hodograph’s estimated storm direction is northeast at 36 knots. The storm that did cross
the entire state of Oklahoma began moving northeast, but turned east early on in its life.
The problem is that with that hodograph, an easterly storm motion would result in almost
no streamwise vorticity being ingested into the updraft within the lowest 1 km. A
northeast moving storm would maximize its streamwise vorticity in the lowest 1 km,
owing to a potentially tornadic storm. Therefore, it seems that the OUN sounding is not
entirely representative, in my opinion, to the actual supercell environment west of
Oklahoma City at that same time. Furthermore, a sounding launched much closer to the
supercell itself would have to be analyzed to really study this particular supercell.
Radar Analysis
While there were over ninety tornadoes on May 29, 2004 across the southern and central
plains, only six supercells were responsible for nearly all of those tornadoes. Though, the
first thunderstorms in the area were mainly elevated thunderstorms forced above the cold
front along the high plains of northeastern Colorado. By 2100 UTC, thunderstorms were
ongoing in western Oklahoma just ahead of the dryline. Two thunderstorms had already
formed along the dryline in north central Kansas, and right on the Kansas/Nebraska
border (Figure 58). Oddly enough, storms were only beginning to initiate in south central
Kansas by 2100 UTC. The weak convergence aloft associated with the left exit region of
the 300 mb jet streak may have strengthened the cap more in this area. Nonetheless, by
2302 UTC (Figure 59), five supercells were present on the VNX radar. The northernmost storm in Kansas, which is visible on the VNX radar, has the characteristics of a low
precipitation supercell. The northern-most storm in Oklahoma does as well. Also, field
observations confirm this storm was a low precipitation supercell. The two furthest south
storms/supercells are within close proximity to each other, which is affecting their
intensity at this time. However, the storm to the north will soon be cut off from the
warm, moist air coming from the south-southeast by the storm to the south, and will soon
dissipate. A very similar result will occur to the storms in southern Kansas. The
supercell furthest south will take all the moisture and CAPE, and the storms nearby to the
north will soon dissipate.
By 2356 UTC, the southern-most storms in both regions have taken over (Figure 60).
The supercell in southern Kansas is located in western Harper County. This storm will be
referred to as storm “B”. With no sign of a hook echo, this supercell has characteristics
of a low precipitation supercell. The storms to its north have weakened to heavy
showers. Meanwhile, the southern-most storm in Oklahoma has taken over in its region
as well. This high precipitation supercell will be referred to as storm “A”.
At 0032 UTC, the TLX (figure 61) radar shows that storm A was an impressive high
precipitation supercell entering northwestern Canadian county, Oklahoma. A
pronounced hook echo can be seen in the southwest portion of the reflectivity pattern.
According to Forbes (1981), a hook echo is the direct result of the mesocyclone
circulation of a supercell. But, the hook is filled in due to the large amounts of
precipitation that is falling near the updraft. A rain wrapped tornado had been reported
only a few minutes prior to this radar image in southeast Blain County near Geary,
Oklahoma. Field observations indicate a very large, rain wrapped low-level mesocyclone
was present at this time. A V-notch is also notable in the higher DBZ values, indicating a
very intense updraft. The 0040 UTC SRV scan from VNX (Figure 62) shows a velocity
couplet associated with the mesocyclone of storm A, which has 50 knot inbound and 50
knot outbound winds. Storm B has not yet developed a strong couplet; with 40-50 knot
inbound winds, but only 10 knot outbound winds.
By 0110 UTC, storm A has a double hook structure (Figure 63), implying the old
mesocyclone has occluded or is in the process of occluding, and the new mesocyclone is
forming downstream (to the east), represented by “New”. A tornado was reported on the
ground at this time. Another feature worth mentioning is the apparent hail spike notable
to the northwest of the core of reflectivity. This is an indicator of very large hail. Field
observations can confirm hail as large as softballs falling in northern Canadian county, on
the southern edge of the hail shaft, jus north of the inflow notch. Eight minutes later, at
0118 UTC (Figure 64), the new hook on radar, which is the radar indication of the new
low-level mesocyclone, has taken over. A pronounced inflow notch is notable just to the
east of the hook echo, indicating very strong winds going into the supercell. Storm A is
now showing signs of slowly evolving into a classic supercell. By 0140 UTC (Figure
65), storm A appears to have more classic supercell radar characteristics, with a skinnier,
and more pronounced hook echo. However, there is still excess reflectivity near the
hook, so this supercell is not a true classic, but it is certainly undergoing the evolution
from high precipitation to classic. At the time of this radar image a large wall cloud was
rapidly rotating on the eastern edge of the hook echo, about to produce a small tornado in
eastern Canadian county. By 0148 UTC, storm A had a very well defined hook echo.
Field observations can confirm a small tornado was on the ground at the time of this radar
image. The tornado emerged from the rain wrapped mesocyclone, moved south, and then
disappeared into the rain wrapped mesocyclone once again. Due to the peculiar motion
of the tornado and its small size, the tornado may have been a satellite tornado rotating
around a large tornado which was wrapped in rain and invisible.
For the next seventy minutes, storm A did not produce a tornado. Instead, it went
through several mesocyclone cycles which were non-tornadic. It continued its eastward
progression through Canadian county, and northern Oklahoma County. At 0239 UTC
(Figure 67), it had an impressive double hook structure once again. Also, storm A had
such a strong Rear Flank Downdraft (RFD), it can be seen on radar as a fine line. Figure
68 shows the base reflectivity evolution of storm A between 0334 UTC and 0407 UTC
from the TLX radar. While at first it appears storm A is undergoing storm splitting, this
may not be the case. Between 0334 UTC and 0338 UTC, it appears either the old
mesocyclone/updraft re-develops, or an anti-cyclonic updraft strengthens. By 0355 UTC,
this mesocyclone has moved northeast, which is aligned pretty well with the shear vector
(based off of the 0000 UTC OUN hodograph). If this was truly a left split updraft, then it
would more than likely propagate toward the north-northeast. Also, this mesocyclone has
a cyclonic hook echo, curing to the southeast. Therefore, it appears this secondary hook
is simply the old mesocyclone back-building at first, then intensifying as it moved off to
the northeast. This mesocyclone did not last much longer after the 0407 UTC as it was
engulfed by the “main” mesocyclone. Also notable on all four radar images is the RFD,
clearly represented by the fine line sloping to the southwest from the “main” hook echo.
Between 0313 UTC and 0651 UTC, several tornadoes were reported by storm A between
Logan county and Mayes county near Tulsa. The supercell finally dissipated as it crossed
the Oklahoma/Arkansas border.
Switching gears back to storm B, we will continue at 0020 UTC, May 30, 2004. The first
tornado from this storm has just been reported in Harper County, Kansas near Anthony.
This was the first of a family of nearly a dozen tornadoes produced by storm B. Not only
that, but this supercell had at least one tornado on the ground (two at times) for the
following two hours. A few minutes after the first tornado report, the 0027 UTC base
reflectivity (Figure 69) from the ITC radar reveals storm B has matured into a spectacular
supercell. Judging from radar alone, the southwester edge of the reflectivity pattern has a
small, rather subtle appendage that could be considered a developing hook. At this time,
it certainly appears like this supercell is some hybrid between classic and low
precipitation. This certainly explains why all the tornadoes were photogenic, with no rain
near the updraft. The “hook” became more evident on base reflectivity by 0044 UTC
(Figure 70), which made this low-precipitation/classic hybrid supercell very impressive
on radar. The small scale yet well defined nature of this particular hook echo is an
indication of the lack of any significant precipitation near the updraft, owing to incredible
visual inspection of the updraft and tornadoes. It may also be an indication of the
intensity of the low-level mesocyclone itself, since the precipitation is being confined to
well a defined, small scale hook rather than a large, obscured hook. Regardless of the
appearance of the subtle hook, this supercell was producing a tornado at those times.
At 0106 UTC, storm B appears to have two mesocyclones (Figure 71), as did storm A
several different times throughout its life. This appears very similar to the storm scale
situation which storm A revealed in Figure 68, where an old updraft/mesocyclone redeveloped or back built behind the “main” updraft. This seems to be the case with storm
B, yet the re-intensifying updraft to the northwest is much weaker, and is likely decaying
at this point. The hook echo associated with the “main” updraft is very classic, owing to
this supercell’s now classic appearance on radar. AT 0132 UTC (Figure 72), storm B
indicated on radar to be evolving into more of a classic/high precipitation supercell. This
is shown by the very thick hook echo with precipitation around and near the hook.
However, by 0145 UTC, the reflectivity that had appeared near the hook just thirteen
minutes earlier is gone. Now there is again a small, tight, well defined hook with no
scattered rain nearby. So, it’s as if storm be was about to undergo the process of evolving
into a high precipitation supercell, but something in the environment kept this from
occurring. Storm B slowly dissipated as it passed through the southern edge of Wichita.
Switching focus again further north, tornadic supercells also tracked along northern
Kansas, northwest Missouri, and central Nebraska. The regional composite reflective
from 2332 UTC (Figure 73) indicates five main supercells, all of which produced a
family of tornadoes. Based off of the 2200 UTC and 0000 UTC surface analysis (Figures
49 and 50, respectively), the central Nebraska storm formed near the intersection of the
dryline/cold front. The north central Kansas supercell formed along the dryline and
warm front, while the northwest Missouri supercell formed along the warm front entirely.
Warmer 700 mb temperatures in southern Kansas and Oklahoma (13 °C), based off of the
0000 UTC 700 mb analysis (Figure 54), kept the storms more isolated in those areas.
The storms were more scattered further north where the 700 mb temperatures were cooler
(10 °C).
Shear Analysis
First, the 0000 UTC OUN sounding’s vertical wind profile indicates approximately 40
knots of 0-6 km shear, which is considered strong shear, and more than enough to support
supercell thunderstorms. The 0000 UTC TOP sounding indicates 43 knots of 0-6 km
shear, which is also considered strong shear. So it is clear, as far as 0-6 km shear is
concerned, why strong supercells existed across Oklahoma and Kansas on May 29, 2004.
However, based on the 0000 UTC OAX sounding, it is not as clear. It indicates 27 knots
of 0-6 km shear, which, based upon what I have researched, falls into the “minimal” or
“marginal” shear category. Therefore, a more detailed analysis is needed.
Based on the 0000 UTC 500 mb analysis, OAX had no wind observation. I find it
difficult to believe the wind was only 10 knots, based on the sounding. Nonetheless, that
is what the data indicates. Even so, the 7 km wind is 30 knots, and given the backed lowlevel flow, this should aid in sufficient vertical shear for supercells. Since there were
several tornadoes reported in northwest Missouri, we need to look at the low-level shear
from the TOP sounding. It indicates a 0-1 km helicity value of 150 m^2/s^2, and a 0-3
km helicity value of 266 m^2/s^2. These values are supportive of tornadoes. Also, the
storm may have routed itself along the warm front, and ingested horizontal vorticity into
the supercell’s updraft, further aiding in low-level mesocyclone generation.
From the 0000 UTC OUN sounding, high 0-1 km and 0-3 km values are indicated, with
264 and 297, respectively. This would certainly support tornadoes, even violent
tornadoes. However, the storm motion indicated by the hodograph is to the northeast.
Both storm A and storm B were moving to the east when they producing most of their
tornadoes. This means the wind profile would have been different to produce a
hodograph which would be conducive for tornadoes with a due east storm motion. Once
again, it does appear that there is something different about the OUN wind profile, which
would not work for the storm motion that occurred on both storm A and B.
A closer inspection of the low-level shear environment can be taken by looking at the
Haviland, Kansas wind profiler (Figure 56). The storm did not become tornadic until just
after 0000 UTC, May 30. Interestingly, the 850 winds increased from 35-40 knots at
2300 UTC to 50 knots at 0000 UTC. They also veered ever so slightly to the southsouthwest. So the speed shear in the 0-1 km (roughly speaking) layer increased by
approximately 10-15 knots. This would certainly increase the 0-1 km storm relative
helicity values, dramatically increasing the tornado potential. It seems the 700 mb winds
are usually 35-40 knots when a significant tornado is produced. In this case, the 700 mb
winds were southwesterly at 40 knots.
As mentioned before, stronger 300mb winds were evident on the 0000 UTC 300 mb
analyses (Figure 53). Norman was right under the 80 knot jet, whereas southern Kansas
had 65 knot 300mb winds. There doesn’t seem to be any other major differences
between the two areas.
Conclusion
On May 29, 2004 a classic tornado outbreak setup unfolded across portions of Oklahoma,
Kansas, Nebraska and Missouri. The ideal thermodynamic and dynamic conditions came
together to produce nearly 90 tornadoes. The two main tornadic supercells that were
focused on were the Harper county, KS storm and the Oklahoma supercell which tracked
across the entire state of Oklahoma.
Storm B had slightly veered 850 mb winds in its local environment, which may have
entrained slightly drier air into the storm. It also formed along a faint dryline bulge,
which may account for weak dry air entrainment near 925 mb, which could cause the
storm to be more classic to low precipitation in nature. It was also north of storm A,
which did not have a storm to its south. Consequently, storm A probably drew in a lot of
the moisture, possibly affecting the moisture availability for storm B; owing to its
appearance. These are only three possible explanations as to why storm B was not high
precipitation, producing photogenic tornadoes throughout nearly its entire life.
Storm A was not under any of the “special” conditions as storm B was. In fact, storm A
was under stronger 300 mb winds, which should have kept more precipitation way from
the updraft, helping it become more classic than high precipitation, which it was for most
of its life. Therefore, I have found the explanations for storm B’s lack of substantial
precipitation near the updraft to be inconclusive. Higher resolution data is needed to
further investigate this case study.
Figure 39a: Preliminary storm reports on May 29, 2004 from the Storm Prediction Center.
Figure 39: 1200 UTC 500 mb analyses from May 29, 2004.
Figure 40: 1200 UTC 300 mb analyses from May 29, 2004.
Figure 41: 1200 UTC 700 mb analyses from May 29, 2004.
Figure 42: 1200 UTC 850 mb analyses from May 29, 2004.
Figure 43a
Figure 43b
Figure 43: Soundings from 1200 UTC May 29, 2004: Left: Figure 43a, Midland, Texas sounding. Right:
Figure 43b, Amarillo, Texas sounding.
Figure 44a
Figure 44b
Figure 44: Soundings from 1200 UTC on May 29, 2004. Left: Figure 44a, Norman, Oklahoma sounding.
Right: Figure 44b, Lamont, Oklahoma sounding.
Figure 45a
Figure 45b
Figure 45: Soundings launched at 1200 UTC on May 29, 2004. Left: Figure 45a, Dodge City, Kansas
sounding. Right: Figure 45b, Omaha, Nebraska sounding.
Figure 46: National surface analysis at 1311 UTC on May 29, 2004.
Figure 47: Surface analysis at 1800 UTC on May 29, 2004. Brown and blue represent the dryline and
cold front, respectively.
Figure 48: Surface analysis at 2100 UTC on May 29, 2004.
Figure 49: Surface analysis at 2200 UTC on May 29, 2004.
Figure 50: Surface analysis at 0000 UTC from May 29, 2004. Notice the wind observation at Gage,
Oklahoma, in northwest Oklahoma is out of the southwest, indicating the dryline has passed through.
Figure 51: National CAPE analysis from the RUC at 0000 UTC May 30, 2004.
Figure 52: 0000 UTC 500 mb analysis on May 30, 2004.
Figure 53: 0000 UTC 300 mb analysis from May 30, 2004.
Figure 54: 0000 UTC 700 mb analysis from May 30, 2004.
Figure 55: 0000 UTC 850 mb analysis from May 30, 2004.
Figure 56: Haviland, Kansas profiler between 0600 UTC May 29, 2004 and 0300 UTC May 30, 2004.
Figure 57: Norman, Oklahoma sounding launched at 0000 UTC May 30, 2004.
Figure 58: Regional base reflectivity composite at 2100 UTC May 29, 2004. Supercell have initiated
along the dryline across western Oklahoma.
Figure 59: VNX (Vance Air Force Base) base reflectivity at 2302 UTC on May 29, 2004.
Figure 60: VNX base reflectivity 2356 May 29, 2004. Two supercells are now evident. One in Oklahoma
(HP) and one in Kansas (LP).
Figure 61: TXL (Twin Lakes, Oklahoma) base reflectivity 0032 UTC May 30, 2004. HP supercell with a
hook echo in Oklahoma.
Figure 62: VNX storm relative velocity tilt 1 0040 UTC May 30, 2004. A strong velocity couplet is
circled on storm A. Storm B has a weaker velocity couplet.
Figure 63: TLX base reflectivity 0110 UTC May 30, 2004. The old and new mesocyclones associated
with storm A are denoted by “New” and “Old”.
Figure 64: TLX base reflectivity 0118 UTC May 30, 2004. Large hook echo on storm A.
Figure 65: TLX base reflectivity 0140 UTC May 30, 2004.
Figure 66: TLX base reflectivity 0148 UTC May 30, 2004. Storm A now has a classic hook echo, and the
storm appears to be more CL than HP on radar. A tornado was observed at this time near the hook.
Figure 67: TLX base reflectivity 0239 UTC May 30, 2004. A double hook structure is noted, denoting the
old and new mesocyclones. The RFD is also noted as a fine line draping southwest from the hook(s).
Figure 68: TLX base reflectivity. Top left: 0034 UTC, two mesocyclones are evident, along with a fine
line associated with the RFD, connecting to the new mesocyclone. Top right: Old mesocyclone is
intensifying. Bottom left: Old mesocyclone is not so “old” anymore, as it has become its own supercell,
splitting from the “main” cell. Bottom right: A cyclonic hook echo is evident on both cells/updrafts.
Figure 69: Wichita, Kansas (ICT) base reflectivity 0027 May 30, 2004. Notice the hybrid LP/CL
appearance of the southern Kansas supercell (storm B).
Figure 70: ICT base reflectivity 0044 UTC May 30, 2004. A tiny hook is noted on storm B.
Figure 71: ICT base reflectivity 0106 UTC May 30, 2004.
Figure 72: ITC base reflectivity. Left: 0132 UTC, notice how thick the hook echo is at this time. Right:
0145 UTC, the hook is very tiny and has consolidated to a small area; very LP/CL on radar.
Figure 73: Regional base reflectivity composite 2332 UTC May 30, 2004. Five main cyclic tornadic
supercells are noted.
Case IV: June 2, 2004 Southern Plains Derecho
On the morning of June 2, 2004 scattered severe thunderstorms formed in west central
Kansas. As more developed, they congealed into a small complex of thunderstorms as
they propagated southeast into central Oklahoma. Once in central Oklahoma, this
thunderstorms complex was a mature squall line, intensifying as it propagated southsoutheast. This squall line intensified and increased in size to be considered a Mesoscale
Convective Complex, making its way into Louisiana and southern Mississippi before
finally weakening. There were approximately 260 damaging wind reports and
approximately 200 large hail reports across Kansas, Oklahoma, Texas, Louisiana,
Mississippi and Alabama (Figure 74a). That is a length greater than 400 km, which by
definition; this high wind event is considered a derecho. This case study will attempt to
explain what caused the initial storms to develop, and why they developed into such a
severe thunderstorms complex.
Synoptic Setup
On the morning of June 2, 2004 at 1200 UTC a 500 mb (Figure 74) neutrally tilted tough
was located over the eastern half of the United States. A deep trough was located over
Hudson Bay, Canada, allowing cold 500 mb temperatures to advect south out of Canada.
This resulted in unseasonably cold 500 mb temperatures over the southern U.S. for early
June (temperatures between -10°C and -15°C). A shortwave trough was located from
central Louisiana north, up the Mississippi River to Tennessee. Modest 35-45 knot
northwesterly flow was present over Oklahoma and Kansas, with temperatures between 14°C and -15°C. There was nothing significant to mention about the 300 mb analysis
over the central/southern plains. There were mainly 55 knot winds from the northwest
across the southern and central plains, however (Figure 75). The right entrance region to
a jet streak was located over Mississippi and Alabama, which was likely aiding the
thunderstorm complex ongoing at that time.
The 700 mb flow across Texas, Oklahoma and Kansas was weak and variable. At 850
mb, there were southeast winds at both Amarillo and Dodge City, implying an upslope
regime was already underway along the high plains (Figure 76). The moisture
availability was ample too, as 15°C and 16°C dewpoints were observed in Norman and
Dallas, respectively. Amarillo reported a dewpoint temperature of 12°C, while Dodge
City reported 12°C.
Based on the 1200 UTC soundings from DDC (Figure 77), and LMN (Figure 78), it is
clear the low levels are still recovering from the frontal passage/outflow boundary
passage which has spread drier air out over the surface across Kansas. This is based on
the comparison of the low level moisture profile from the 1200 UTC OUN sounding
(Figure 79), which shows a rich layer of moisture from the surface to just above 900 mb.
This is somewhat shallow, which suggests the surface front/outflow boundary moved
north overnight. In any case, the DDC sounding has a frontal inversion from the surface
to 850 mb with a corresponding backing of the winds, and an 800 mb dry layer.
Beginning above the cold pool, the wind profile veers more than 90 degrees over 6 km,
and shows 45 knots of 0-6 km shear. This would support supercell thunderstorms;
however, there is no surface based CAPE. The sounding does indicate 664 J/kg of ML
CAPE. Based on the above analysis, it appears the initial thunderstorms formed just
above the frontal inversion at 850 mb. That being the case, these storms were likely
elevated supercells (Figures 80a and 80b).
The 1200 UTC LMN sounding (Figure 78) also has a frontal inversion, with a
corresponding backing of the winds, located from the surface to about 925 mb. A mixed
layer is located just above the frontal inversion, indicating the rich boundary layer which
was in place prior to the frontal passage. A more pronounced dry slot, more so than the
DDC sounding, and a corresponding elevated mixed layer, is located between 825 mb
and 800 mb. Once again there is no surface based CAPE, yet the wind profile beginning
above the inversion will support supercells, with more than 90 degrees of direction shear
and 45 knots of 0-6 km shear. There is 1398 J/kg of ML CAPE, which will support
elevated supercells. Based on all the information provided from this sounding, the storms
which initiated northwest of Dodge City (DDC) and moved southeast would be able to
strengthen as they moved into an environment with more ML CAPE, and eventually into
a richer moisture environment.
This “richer” moisture environment is seen from the 1200 UTC OUN sounding south of
the baroclinic zone (Figure 79). Again, the rich moisture layer is shallow. But, the winds
between the top of the moisture layer and 850 mb are from the southeast, advecting
moisture into the area. Surface based CAPE is 818 J/kg, which will support surface
based convection. While there is approximately 40 knots of 0-6 km shear and more than
90 degrees of directional shear, the winds in the midlevels (near 700 mb) are very weak.
Also, almost all the speed shear is confined above 3 km. This shear profile will not
support supercells, but will support multi-cells and severe linear storms.
The 1200 UTC surface map (Figure 81c) reveals relatively low dewpoints in the 40’s°F
across southern Kansas. But further south in Oklahoma and Texas surface dewpoints are
in the 60’s°F. As such, a stalled out frontal boundary or outflow boundary (which can
safely be called a baroclinic zone) was oriented generally east-west across the
Kansas/Oklahoma border. This will become more diffuse as the day progresses and solar
radiation warms the boundary layer. The composite radar shows several weak echo lines
across Kansas, probably suggesting small scale outflow boundaries. This, along with the
rather variable surface winds, suggests to me outflow boundaries existed. Therefore, it
seems likely the interaction between outflow boundaries, upslope flow in the lower
levels, isentropic assent along and behind the front, and boundary layer mixing caused
the first several thunderstorms to initiate around 1200 UTC (Figure 80a 1233 UTC and
80b 1356 UTC). This goes well with what (Johns and Hirt, 1986) found, that derechos
are almost always associated with a quasi-stationary thermal boundary along which warm
air advection develops.
Radar Analysis I: Derecho Development
From the ICT 1356 UTC radar image (Figure 80b); there is a strong cell just east of
Dodge City. Since it is so far away from the ICT radar, this image represents the
precipitation pattern in the upper levels. However, it is a single cell with a 60 DBZ core,
which could have been an elevated supercell. Other, more disorganized convection has
begun along the frontal boundary east of Dodge City and southeast of Wichita. Weak
showers are beginning to form north of the boundary in central Kansas. By 1403
UTC/1459 UTC (Figures 81 and 82), the thunderstorms around and north of Dodge City
are becoming more widespread and are intensifying. Showers are also developing across
northwestern Oklahoma, which is likely a result of gradual warm air advection along or
just ahead of the front. It is clear that, overall, most of the convection is elevated, a result
of isentropic lift above and behind the frontal boundary, along with outflow boundary
interactions. Again, Johns and Hirt (1986) mention that this quasi-stationary surface
boundary plays a significant role in most derecho cases. It is certainly playing a
significant role in this case as well.
By 1602 UTC a cluster of severe thunderstorms and rain was located over south central
Kansas (Figure 83). About 45 minutes later (Figure 84), the rain and severe
thunderstorms had congealed into a thunderstorm complex, located just south of Wichita.
By 1856 UTC, a west-southwest to east-northeast line of severe storms extended across
north central Oklahoma (Figure 85). The southern-most storm did have an inflow notch
and a small hook-like appendage. The base velocity does show a small couplet, 15 knots
outbound and 15 knots inbound, right where the appendage is on the base reflectivity. At
1959 UTC, the squall line had made its way into the Oklahoma City area (Figure 86).
The outflow boundary associated with the cold outflow from the squall line can be seen
as a fine line on radar, oriented east-west through Oklahoma City.
The 1800 UTC surface map shows 10-15 knot easterly winds in the Texas panhandle, and
10 knot easterly winds in southeastern Colorado (Figure 87). This upslope regime is
aiding in thunderstorm development, which began at 2004 UTC in the Texas panhandle
and along the high terrain of southeastern Colorado (Figure 88). The squall line in
Oklahoma appears to be a Mesoscale Convective Complex (MCC), with heavy strataform
precipitation occurring behind the line. Abundant moisture is available ahead of the
MCC with middle 60°F to middle 70°F degree dewpoints present across southern
Oklahoma, Texas and Louisiana.
By 2100 UTC, the MCC across Oklahoma now stretched into central Arkansas (Figure
89). Further west in the Texas panhandle, isolated storms were ongoing, in addition to a
new cluster of thunderstorms developing near Lubbock. The 2100 UTC surface map
(Figure 90) indicates a dryline in west Texas which has mixed eastward from eastern
New Mexico. The storms are aligned in a north/south oriented bulge, much like the
dryline at that time. So, it is clear some mixture of upslope flow and convergence along
the dryline has caused these thunderstorms to initiate. Thunderstorms were also
becoming more widespread across southeastern Colorado, and more storms were forming
over central Kansas. A close look at the MCC in central Oklahoma (Figure 91) indicates
a severe line of thunderstorms.
The 2231 UTC regional radar composite show several clusters of convection across the
southern plains (Figure 92). The MCC is now a severe line draped from southern
Oklahoma east into Arkansas, with heavy precipitation behind the squall line.
Thunderstorms have formed all along the dryline in west Texas, and more have fired
along the outflow boundary in the Texas panhandle. By 0002 UTC these thunderstorms
have clustered together to form a classic comma shaped MCC, which is about to merge
with the older MCC in Oklahoma and Arkansas. This older MCC is now a strong squall
line, with continual heavy strataform precipitation behind the heaviest part of the line.
A closer inspection of the leading edge of the squall line near Fort Worth, Texas (Figure
93) indicates the line was not nearly as intense further west than it was further east in
eastern Texas into Louisiana. In fact, it can clearly be seen that between 2356 UTC and
0057 UTC, the line weakens considerably over north central Texas (Figure 93a and 93b).
On the KFW radar at 2356 UTC (Figure 93a), a fine line can be seen just ahead of the
leading edge of the squall line, indicating the thunderstorm’s outflow. By 0057 UTC
(Figure 93b), this fine line has pushed out ahead of the squall line by a considerable
distance. The leading edge of the squall line is much weaker with the heaviest
precipitation just behind the leading edge, indicating the line is weakening. However,
from the 0002 UTC regional radar composite (Figure 92), the squall line had a long bow
shape over the Arklatex region, which would indicate it is where the strongest part of the
squall line is located. Therefore, a closer inspection of that region is necessary.
Looking at the KSHV 2356 UTC radar image, an impressive squall line can be seen over
Texas and Arkansas (Figure 93a). Unlike the western end of the line in north central
Texas, the eastern part of the line has individual cells clustered together, rather than an
almost continuous line of convection like the western end. This makes the eastern end
appear more severe, possibly containing larger hail and more damaging straight line
winds. At 0057 UTC (Figure 93b), the squall line is very much bowing to the southeast.
One area of particularly intense rear inflow jet winds is located where the base
reflectivity is much lower than the surroundings. This hole of reflectivity is a radar
indication of a rear inflow jet, caused by dry air entrainment in the middle levels of the
atmosphere, resulting in the evaporation of rain in the downdraft of the thunderstorm, and
a downburst of drier air. This is supported by Johns and Hirt (1986); and Doswell III and
Evans (2001). The KSHV 0057 UTC base velocity image (Figure 93c) verifies this claim
with a large swath of 36-50 knot winds coming toward the radar over extreme
southwestern Arkansas and extreme northeastern Texas. There is even a small area in
which the winds are greater than 64 knots, indicated by the purple and pink, where the
values are being range folded.
Shear and CAPE Analysis
The 0000 UTC upper air observations (Figure 95) don’t reveal anything significant about
the environment that wasn’t already known at 1200 UTC. Although, the 850 mb cold
front (Figure 95d) is better defined, practically outlining the Red River, and then
stretching further west into the Texas panhandle. It’s not surprising the largest
temperature gradient is located over the Texas panhandle and west Texas, where the
dryline has mixed east, causing compressional heating. The 0000 UTC FWD sounding
(Figure 96) indicates a deep moisture profile, a weak capping inversion, and a substantial
dry layer at 700 mb. This dry layer should aid in downdraft potential or rear inflow jets,
which in turn should aid in the development of severe bow echoes within the squall line.
Also, there is 4735 J/kg of observed surface based CAPE. That is a substantial amount of
CAPE, which will keep this squall line severe for much longer. Interestingly, the area in
which the strong rear inflow jet winds are located is the same area where there is not a
significant dry slot at 700 mb. While the 0000 UTC Shreveport (SHV) sounding does
have a dryer layer at 700 mb (Figure 97), it is not as significant as the FWD sounding.
The 700 mb and 500 mb winds are also slightly stronger in magnitude on the FWD
sounding as compared to the SHV sounding. The SHV sounding does indicate 3274 J/kg
of surface base CAPE. An analysis of the 0000 UTC LCH sounding (Lake Charles,
Louisiana Figure 98) reveals 4285 J/kg of surface based CAPE. So there is more than
enough CAPE to keep this squall line going all the way south to the Gulf Coast. But,
middle and upper level winds weaken substantially that far south. Furthermore, it seems
the system’s outflow will increase in intensity somewhere between Shreveport and Lake
Charles, and then slowly decrease in intensity as the outflow propagates away from the
squall line.
Radar Analysis II: Derecho Evolution
A glance at the KSHV 0158 UTC base reflectivity image (Figure 99) indicates a fine line,
or outflow boundary, which has propagated ahead of the squall line itself. This has
occurred after passing through Shreveport, where the edge of the stronger middle level
winds are likely located. Further northeast, where the middle level winds are stronger,
the squall line is more intense, having higher DBZ values. Figure 100 shows the regional
radar. The line of severe thunderstorms which began in west Texas is now a MCC
merging with the western end of the older MCC. By 0259 UTC, the KSHV radar (Figure
101) indicates the squall line has actually re-intensified south of Shreveport. The highest
DBZ values are still located along the northern-most part of the line. However, the fact
remains the line did intensify, so the middle level winds must be sufficient in the region.
The base velocity image (Figure 102) indicates the squall line is producing 50-64 knot
winds across a large portion of the squall line area. The swath begins where the rear
inflow jet would be located, and ends near the leading of the squall line. There are also a
couple areas just ahead of the squall line in which the radar signal is being range folded.
So the best estimate I can make on the wind speed is greater than 64 knots. Looking back
at the FWS radar at 0156 UTC (Figure 103) the squall line has lost much of its intensity.
However, a well defined fine line is seen moving through the Dallas/Fort Worth area,
causing straight line wind damage. There are 3 discrete severe thunderstorms over
western north Texas, which are associated with the new MCC which originally developed
over west Texas.
At 0331Z (Figure 103), the regional radar composite shows two mesoscale convective
complexes. One is over western north Texas and the western half of Oklahoma, and the
other over eastern Texas, Arkansas and Louisiana. Both MCCs appear to be intense, for
the most part. But by 0433 UTC (Figure 104), the western MCC has weakened
significantly to a complex of heavy rain. It is not surprising that it weakened. The cold
outflow from the MCC to its east likely aided in the demise of the western MCC, along
with cutting off its access to warm, moist air.
By 0603 UTC, the MCC had made it to Mississippi. Six hours later the MCC was
weakening, but still ongoing over portions of Mississippi, Alabama and Louisiana.
Conclusion
On June 2, 2004 two Mesoscale Convective Complexes formed over the southern plains
affecting Kansas, Oklahoma, Arkansas, Louisiana, Mississippi and Alabama. This
caused approximately 260 damaging wind reports and 200 large hail reports over a path
length greater than 400 km, allowing this to be considered a derecho event. The most
concentrated area of wind damage was located in an area where Oklahoma, Texas,
Arkansas, and Louisiana intersect, sometimes called Arklatex.
The conditions which caused the longevity of this convective complex are not
understood, nor has it been explained in general throughout the meteorology literature
(Weisman, 1992). However, the derecho did originate north of a quasi-stationary thermal
surface boundary, and was aided strongly by warm air advection in the low-levels, as
mentioned in the literature by (Johns and Hirt, 1986). The event also occurred under a
northwesterly flow regime at 500 mb, which is yet another condition Johns and Hirt
mentioned as a likely scenario for derecho events. Furthermore, even though the
southern plains and parts of the Gulf coast states are not favored climatologically for
derecho events, this particular derecho event formed under very similar conditions to
those which are climatologically favored areas.
Figure 74a: Preliminary storm reports for June 2, 2004 from the Storm Prediction Center.
Figure 74: 1200 UTC 500 mb analyses June 2, 2004.
Figure 75: 1200 UTC 300 mb analyses June 2, 2004.
Figure 76: 1200 UTC 850 mb analyses June 2, 2004.
Figure 77: Dodge City, Kansas sounding 1200 UTC June 2, 2004. A shallow cold pool is evident with the
temperature inversion and northeasterly wind at the surface.
Figure 78: Lamont, Oklahoma sound 1200 UTC June 2, 2004. A shallow cold pool is also evident here.
Figure 79: Norman, Oklahoma sounding 1200 UTC June 2, 2004.
Figure 80a
Figure 80b
Figure 80: Left: Figure 80a, regional base reflectivity composite 1233 UTC June 2, 2004. Outflow
boundaries can be seen, along with developing thunderstorms across western and southern Kansas. Right:
Figure 80b, ITC base reflectivity 1256 UTC June 2, 2004. An elevated thunderstorm can be seen east of
Dodge City, possibly an elevated supercell.
Figure 81c: Surface plots 1200 UTC June 2, 2004.
Figure 81: Regional base reflectivity composite 1403 UTC June 2, 2004.
Figure 82: ICT base reflectivity 1459 UTC June 2, 2004.
Figure 83: Regional base reflectivity composite June 2, 2004.
Figure 84: ICT base reflectivity 1648 UTC June 2, 2004.
Figure 85: TLX base reflectivity 1856 UTC June 2, 2004.
Figure 86: TLX base reflectivity 1959 UTC June 2, 2004. Thunderstorm outflow (gust front) is seen as a
fine line oriented east/west across Oklahoma City.
Figure 87: Surface plots 1800 UTC June 2, 2004.
Figure 88: Base reflectivity composite 2004 UTC June 2, 2004. Thunderstorms are initiated in the Texas
panhandle and west Texas along the dryline, and storms are firing from upslope flow across southeast
Colorado.
Figure 89: Base reflectivity regional composite 2101 UTC June 2, 2004.
Figure 90: TLX base reflectivity 2159 UTC June 2, 2004.
Figure 91: Base reflectivity regional composite 2231 UTC June 2, 2004.
Figure 92: Base reflectivity composite 0002 UTC June 3, 2004. Thunderstorms in west Texas have
organized into a MCC. The MCC along the Red River is becoming a strong bow echo.
Figure 93a: FWS base reflectivity 2356 UTC June 2, 2004. Notice how the shall line is relatively solid
and has high DBZ values.
Figure 93b: Fort Worth (FWS) base reflectivity 0057 UTC June 3, 2004. The thunderstorms within the
line have collapsed. An outflow boundary is indicated by the fine line oriented northwest/southeast across
the northern side of Dallas/Fort Worth.
Figure 94a: Shreveport, Louisiana (SHV) base reflectivity 2356 UTC June 2, 2004. The squall line is
shaped like a bow, indicating this part of the line is a bow echo, which likely contains strong straight line
winds.
Figure 94b: SHV base reflectivity 0057 UTC June 3, 2004. A radar indication of a rear inflow jet is
denoted by the precipitation hole. This is an area where dry air entrainment has resulted in evaporation,
and likely a strong downburst.
Figure 94c: SHV storm relative velocity tilt 1 0057 UTC June 3, 2004. Greater than 64 knot winds out of
the north or northwest measured where the precipitation hole and rear inflow jet were noted in Figure 94b.
Figure 95: All 0000 UTC June 3, 2004. Top left: 300 mb. Top right: 500 mb. Bottom left: 700 mb.
Bottom right: 850 mb.
Figure 96: Fort Worth sounding 0000 UTC June 3, 2004. Note the capping inversion at 850 mb. This
kept isolated, surface based, possibly supercellular convection from forming ahead of the squall line.
Figure 97: Shreveport, Louisiana sounding 0000 UTC June 3, 2004. Note 3274 J/kg of CAPE.
Figure 98: Lace Charles, Louisiana sounding 0000 UTC June 3, 2004. Note the 4285 J/kg of CAPE.
Figure 99: SHV bas reflectivity 0158 UTC June 3, 2004. An outflow can be seen as a fine line.
Figure 100: Regional base reflectivity composite 0201 UTC June 3, 2004. The two MCCs are merging.
Figure 101: SHV base reflectivity 0259 UTC June 3, 2004. A large area behind the leading edge of the
squall line has nearly free of precipitation. This may be an indication of a strong downburst.
Figure 102
Figure 102: SHV storm relative velocity tilt 1 0259 UTC June 3, 2004. There are strong winds in parts of
the nearly precipitation free area, but nothing significant.
Figure 103: FWS base reflectivity 0156 UTC June 3, 2004. A well define outflow boundary is moving
through Dallas/Fort Worth.
Figure 104: Regional base reflectivity composite 0331 UTC June 3, 2004. Two strong MCCs are
ongoing. The western-most MCC will soon dissipate since the eastern-most MCC is cutting off its
moisture supply.
Figure 105: Base reflectivity composite 0433 UTC June 3, 2004. The western-most MCC is dissipating as
mentioned in Figure 104.
Figure 106: Base reflectivity composite 0603 UTC June 3, 2004.
IV. Discussion
The four cases discussed in this paper, May 12, May 22, May 29, and June 2, 2004, were
all significant severe weather outbreaks across the central U.S. May 12, May 22, and
May 29 were all significant tornado events, while June 2 was a derecho event.
Nonetheless, all these cases have some fundamental similarities regarding specific
thermodynamic profiles, while they all have major differences as well.
Thermodynamics
All four cases involved a deep layer of rich low-level moisture. In fact, they all had 850
mb dewpoint temperatures of at least 15°C, which is considered tropical moisture. The
actual surface dewpoint temperatures did have some variation, however. On May 12,
surface dewpoints were in the middle 60’s to lower 70’s (°F) across southern Kansas and
northern Oklahoma. The surface dewpoints on May 22 were in the mid and upper 60’s,
while Oklahoma, Kansas and Nebraska saw dewpoints in the lower 70’s on May 29.
During the June 2 event, the derecho was supplied with dewpoints in the 60’s (all in °F).
CAPE values were also relatively similar between all four cases, with 3000 J/kg of CAPE
existing in all four cases. The only difference is on June 2, the original thunderstorm
development only had approximately 1300 J/kg of CAPE initially, but moved into an area
of approximately 3000 J/kg of CAPE.
Shear Profiles
Interesting, all four cases had 0-6 km shear of at least 40 knots. But May 12, 22, and 29
were all associated with significant tornadoes, while June 2 was a derecho event. Also,
on May 12, 22, and 29, the 700 mb winds were at least 40 knots, while the 700 mb winds
on June 2 were very weak. But, as what has already been stated earlier in the paper,
shear profiles do not definitively determine storm mode. The combination of the shear
profile, synoptic scale forcing, CAPE, and the orientation of the shear vector to the
surface boundary all go into consideration of storm mode.
In the case of June 2, 2004, the initial storms in Kansas were not surface based, and were
aided almost entirely by war air advection above the capping inversion. There were one
or two supercells in the Texas panhandle, but they were quickly engulfed by the second
MCC. Furthermore, the environment was supportive of supercells across Oklahoma and
north Texas. However, the capping inversion kept isolated cells from forming ahead of
the squall line.
Out of data collected from May 12, 22, and 29, it seems a 700 mb wind of 40 knots or
greater is necessary to support significant tornadoes. This claim is obviously biased to
these three cases, but will hopefully be used in future tornado forecasts to better evaluate
its accuracy.
Synoptic Setup
June 2 had a completely different synoptic setup than any of the other cases, which is one
good reason why it was a high wind event, and not a tornado outbreak. May 12 was
characterized by little to no upper level support. Supercells initiated along a dryline,
much like they did on May 29 across Oklahoma Kansas. However, moderate to strong
upper level support was present during the May 29 case. Moderate upper level support
was also present on the May 22 case. Furthermore, the more classic tornado outbreaks
seem to have some upper level support present, but that does not mean it is a necessary
condition for a tornado outbreak to occur. Strong midlevel capping inversions were
present on May 12, 22, and 29, especially in the areas where cyclic tornadic supercells
occurred. A capping inversion was present on June 2 as well, which was strong enough
across Oklahoma and Texas to keep isolated convection from forming ahead of the squall
line.
Many different mechanisms caused supercell thunderstorm initiation during the tornado
cases. The dryline initiated supercells in southern Kansas on May 12. While the dryline
did initiate one supercell in northern Kansas on May 22, the majority of the supercells
across Nebraska and Iowa formed along and north of the warm front. Although, the most
significant supercell formed along a dryline/outflow boundary intersection. Also,
orographic lift from an upslope regime in northeastern Colorado and western Nebraska
helped to initiate the supercells in that area. On May 29, the dryline was the main
initiator across the southern plains, while tornadic supercells also formed along the warm
front in northern Missouri and Nebraska. While storms initiated north of a baroclinic
zone on June 2, unlike the May 22 situation, the storms were elevated.
V. Conclusion
This paper diagnosed the conditions on four separate severe weather outbreaks which
occurred in the spring of 2004: May 12, May 22, May 29, and June 2. The first three
were tornado outbreaks, and the last was a derecho event, respectively. May 22 and May
29 were major tornado outbreaks, each having over 70 tornadoes reported during the
respected event. Some interesting characteristics of each event were revealed.
Unfortunately, more specific answers about tornado size and/or intensity were not able to
be definitely answered.
Probably the most interesting aspect of these cases as a whole is their uniqueness to one
another. None of the events were the same, yet similar atmospheric conditions existed in
each case.
Like many other case studies, a higher resolution of upper air data would be necessary to
draw more conclusions. Answers to why the Wilber/Hallam, Nebraska tornado grew to
an unprecedented 2.5 miles wide are simply not here. Educated guesses and theories
were offered, but the study did not uncover the answer. As for the storm A and B on
May 29, little was revealed from the data collected. I have offered likely theories, but I
still do not know why storm B was a LP/CL hybrid supercell, while storm A was a HP
supercell during most of its life. A radiosonde observation from both storm environments
would probably provide answers to these questions, yet that data does not exist. At any
rate, this paper will hopefully be used as a tool to improve severe convective forecasting
in the future.
References
aufm Kampe, H. J., 1960: Tornadoes and mountain-wave effect. J. Meteor., 17, 89-91.
Beeb, B. B., and F. C. Bates, 1955: A mechanism for assisting in the release of
convective instability. Mon. Wea. Rev., 83, 1-10.
___, Doswell, C. A, and Wilhelmson, R. B., 1993: The Role of Midtropospheric Winds
in the Evolution and Maintenance of Low-Level Mesocyclones. Mon. Wea. Rev.,
122, 126-134.
Bluestein, H. B., Hane, C. E. and Ziegler, C. L., 1993: Investigation of the Dryline and
Convective Storms Initiated along the Dryline: Field Experiments during COPS91. Bul. Amer. Met. Soc, 74, No. 11, 2136.
___, McCaul, E. W. Jr., Byrd, G. P., Woodall, G. R., 1988: Mobile Sounding
Observations of a Tornadic Storm near the Dryline: The Canadian, Texas Storm
of 7 May 1986. Mon. Wea. Rev, 116, 1790-1802.
___, Thomas, K. W., 1984: Diagnosis of a Jet Streak in the Vicinity of a Severe Weather
Outbreak in the Texas Panhandle. Mon. Wea. Rev., 112, 2499-2504.
___, Weiss, C. C., Pazmany, A. L., 2003: Mobile Doppler Radar Observations of a
Tornado in a supercell near Bassett, Nebraska, on 5 June 1999. Part I:
Tornadogenesis. Mon. Wea. Rev. 131, 2954-2959.
Brooks, H. E., Doswell III, C. A., and Cooper, J., 1994: On the Environments of
Tornadic and Nontornadic Mesocyclones. Weather and Forecasting, 9, 606-616.
___, C. A. Doswell III., and R. Davies-Jones, 1993: Environmental helicity and the
maintenance and evolution of low-level mesocyclones. The Tornado: Its
Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., C. Church,
Ed., No.79, Amer. Geophys. Union, 97-104.
Browning, K. A., 1964: Airflow and Precipitation Trajectories Within Severe Local
Storms Which Travel to the Right of the Winds. J. Atmos. Sci., 21, 634-636.
___, and F. H. Ludlam, 1962: Airflow in convective storms. Quart. J. Roy. Meteor. Soc.,
88, 117-135.
Danielsen, E., 1974: The relationship between severe weather, major dust storms and
rapid large-scale cyclogenesis. Parts I and II. Subsynoptic Extratroptical Weather
Systems, M. Shapiro, Ed., 215-241.
Davies, J. M., Johns, R. H., 1993: Some Wind and instability Parameters Associated
With Strong and Violent Tornadoes. 1. Wind Shear and Helicity. The Tornado:
Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., C. Church,
Ed., No. 79, Amer. Geophys. Union, 573-581.
Davies-Jones, R., and H. Brooks, 1993: Mesocyclogenesis from a theoretical
perspective. The Tornado: Its Structure, Dynamics, Prediction, and Hazards,
Geophys. Monogr., C. Church, Ed., No. 79, Amer. Geophys. Union, 105-114.
Dowell, D. C., Bluestein, H. B., 2002: The 8 June 1995 McLean, Texas, Storm. Part II:
Cyclic Tornado Formation, Maintenance, and Dissipation. Mon. Wea. Rev., 130,
2649-2653.
Doswell III, C. A. and Burgess, D. W., 1993: Tornadoes and Tornadic Storms: A
Review of Conceptual Models. The Tornado: Its Structure, Dynamics,
Prediction, and Hazards, Geophys. Monogr., C. Church, Ed., No. 79, Amer.
Geophys. Union, 161-172.
Doswell III, C. A., 1991: A review for forecasters on the application of hodographs to
forecasting severe thunderstorms, Natl. Weather Dig., 16, 2-16.
___, C. A and Evans, J. S., 2001: Examination of Derecho Environments Using
Proximity Soundings. Weather and Forecasting, 16, 329-342.
Droegemeier, K. K., and R. B. Wilhelmson, 1985: Three-dimensional numerical
modeling of convection produced by interacting thunderstorm outflows. Part I:
Control simulation and low-level moisture variations. J. Atmos. Sci., 42, 23812403.
Fawbush, E. J., and R. C. Miller 1952: A mean sounding representative of the tornadic
airmass environment. Bull. Amer. Meteor. Soc., 33, 303-307.
Foote, G. B., 1977: Response to “The structure and mechanism of hailstorms”, Meteor.
Monogr., No. 38, 45-47.
Forbes, G. S. 1981: On the reliability of hook echoes as tornado indicators, Mon.
Weather Rev., 109, 1457-1466.
Hookings, G. A., 1965: Precipitation-maintained downdrafts. J. Appl. Meteor., 4, 190195.
___, and R. C. Miller, 1954: The types of airmasses in which North American tornadoes
form. Bull. Amer. Meteor. Soc., 35, 154-165.
___, and L. G. Starrett, 1951: An empirical method for forecasting tornado development.
Bull. Amer. Meteor. Soc, 32, 1-9.
Johns, R. H., 1983: A Synoptically Climatology of Northwest-Flow Severe Weather
Outbreaks. Part II: Meteorological Parameters and Synoptic Patterns. Mon. Wea.
Rev., 112, 449-462.
___, and Hirt, W. D., 1986: Derechos: Widespread Convectively Induced Windstorms*.
Weather and Forecasting, 2, 32-49.
Klemp, J. B., and R. B. Wilhelmson, Simulations of right-and left-moving storms
produced through storm splitting, J. Atmos. Sci. 35, 1097-1110, 1978b.
Maddox and Doswell, 1982: Significant Tornadoes 1680-1991, Environmental Films, T.
P. Grazulis, E., 85.
Maddox, R. A., L. R. Hoxit, and C. F. Chappell, 1980: A study of tornadic thunderstorm
interactions with thermal boundaries. Mon. Wea. Rev., 110, 184-197.
Moller, A. R., Doswell III, C. A., Foster, M. P., Woodall, G. R., 1994: The Operational
Recognition of Supercell Thunderstorm Environments and Storm Structures.
Weather and Forecasting, 9, 327-345.
Newton, C. W., 1963: Dynamics of severe convective storms. Meteor. Monogr., 5, No.
27, 33-58.
Rasmussen, E. N., and R. B. Wilhelmson, 1983: Relationships between storm
characteristics and 1200 GMT hodographs, low level shear, and stability.
Preprints, 13th Conf. on Severe Local Storms, Tulsa, OK, Amer. Meteor. Soc., J5J8.
Roebber, P. J., 1984: Statistical analysis and updated climatology of explosive cyclones.
Mon. Wea. Rev., 11, 1577-1589.
Rotunno, R., 1993: Supercell Thunderstorm Modeling and Theory. The Tornado: Its
Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., C. Church,
Ed., No. 79, Amer. Geophys. Union, 57-73.
___, and J. B. Klemp, 1985: On the rotation and propagation of simulated supercell
thunderstorms. J. Atmos. Sci., 42, 271-292.
___, and J. B. Klemp, 1982: The influence of the shear-induced pressure gradient on
thunderstorm motion. Mon. Wea. Rev., 110, 136-151.
Schaefer, J. T, 1986: Severe Thunderstorm Forecasting: A Historical Perspective.
Weather and Forecasting, 1, 164-184.
Shuman, F. G., and L. P. Carstensen 1952: A preliminary tornado forecasting system for
Kansas and Nebraska. Mon. Wea. Rev., 80, 233-240.
Srivastava, R. C., 1985: A simple model of evaporatively driven downdraft: Application
to microburst downdraft. J. Atmos. Sci., 42, 1004-1023.
Thorpe, A. J., M. J. Miller, and M. W. Moncrieff, 1982: Two-dimensional convection in
non-constant shear: A model of midlatitude squall lines. Quart. J. Roy. Meteor.
Soc., 108, 739-762.
Tibaldi, S., A. Buzzi, and A. Speranza, 1990: Orographic cyclogenesis. Extratropical
Cyclones, C. Newton and E. O. Holopainen, Eds., Amer. Meteor. Soc., 107-127.
Ucellini, L. W., and D. R. Johnson, 1979: The coupling of upper and lower tropospheric
jet streaks and implications for the development of severe convective storms.
Mon. Wea. Rev., 107, 682-703.
Weisman, M. L., 1992: The Genesis of Severe, Long-Lived Bow Echoes. J. Atmos. Sci.,
50, No. 4, 645-670.
___ and Klemp, J. B., 1982: The Dependence of Numerically Simulated Convective
Storms on Vertical Wind Shear and Buoyancy. Mon. Wea. Rev., 110, 504-519.
___ and Rotunno, R., 2003: “A Theory for Strong Long-Lived Squall Lines” Revisited.
J. Atmos. Sci, 61, No. 4, 361-381.
Wicker, L. J., and R. B. Wilhelmson, 1995: Simulation and analysis of tornado
development and decay within a three-dimensional supercell thunderstorm. J.
Atmos. Sci., 52, 2675-2703.
Wilhelmson, R. B., and J. B. Klemp, 1978: A numerical study of storm splitting that
leads to long lived storms. J. Atmos. Sci., 35, 1974-1986.
Download