Title: Cell Initiation, Interaction, and Morphology in the 1996 Illinois Tornado Outbreak
1Cell Initiation, Interaction, and Morphology in
the 1996 Illinois Tornado Outbreak
Brian F. Jewett and Robert B. Wilhelmson U.I.
Atmospheric Sciences NCSA
Bruce D. Lee University of Northern Colorado
2April 19, 1996 Damage
- Most Illinois tornadoes in a single day (36)
- Average IL tornadoes in a year 27
- Tornadoes hit Jacksonville, Bloomington,
Decatur, Champaign-Urbana, Ogden - 1 fatality in IL (I-74 near Ogden)
3April 19, 1996 Motivation
- Many tornadoes - but
- Few(er) tornadic storms
- Large number of initial storms fewer mature
storms later
- Most tornadoes short-lived
- Some tornadoes up to low F-3 intensity (C-U
Ogden)
4April 19, 1996 Motivation
- Frequent instances of cell splitting and merging
- Result was fewer, but tornadic, storms moving
across IL - Tornadogenesis found to often accompany later
mergers
Cell evolution in western Illinois
5Methodology MM5
- Cyclone evolution MCS simulated using MM5
- Up to 5 domains (1 km grid)
- Start simulation at 00z run 27h
Inner 4 model domains
- Mesoscale evolution reasonable
- MM5 results also exhibit cell splitting and
merging - Split/merge details strongly dependent on
resolution
1-km rainwater at 2236, looking N
6Convective Initiation
- Model storms formed correct time (aftn)
location (e. Missouri) - Storms formed on, moved off surface
dryline - Convection appeared to remain near leading
edge of 600 mb qe gradient - CFA?
7Convective Initiation
- Cross section at mature time shows apparent
cold front aloft - Earlier fields less clear qe gradient due
almost entirely to dry air - Convection fires in model as leading qe
gradient moves over surface drytrough - Coincidence/causality?
qe (red contours) rainwater (green outline,
shading)
8Idealized Studies
- Idealized 3-D simulations were made with WRF to
investigate cell interaction without - pre-existing boundaries
- changes in instability/shear across the domain
- changes in instability/shear over time
We explored the development mature structure of
a pair of storm cells, varying only their
original position. Questions what types of
interaction are favorable? How sensitive is storm
evolution to cell orientation?
9Methodology WRF
- Soundings extracted from MM5 simulation at one
of three locations on dryline. - Location / time chosen where convection
formed in the MM5 simulation. - WRF runs
- Sounding A
- 1 km grid spacing
- 2.5 hour simulations
- horizontally uniform
hodograph from location A
10Methodology WRF
- 230 WRF runs were made, two control
(single-cell) - Each corresponded to a particular
arrangement of a pair of initial storm cells - In figure at left
- Each square 1 simulation
- 1st storm in the middle
- 2nd at one of blue squares
- Center cell stronger
Matrix of WRF simulations
11Rotational properties
Data was saved each minute, for 2.5 hours.
Statistics gathered
- Max 1 km, overall updraft
- Peak surface wind speed
- Peak surface vorticity
- Max rotation duration
Magnitude
Duration
Most cases of interaction were destructive
cells stronger w/o neighbor
12Rotational properties
Vorticity properties (a single number) for each
run were plotted in the same coordinates as the
initial run matrix, showing the relationship
between surface-based rotation initial cell
layout.
Magnitude
Duration
Run matrix
13Rotational Magnitude
Of interest was the peak surface vorticity when
the secondary cell formed to the southwest of the
stronger, center storm. The opposite was also
true - weaker rotation resulted when the 2nd cell
formed to the northeast. A nearly N-S
orientation led to weaker rotation.
14Rotational Duration
Most surprising here was the very sharp gradient
at the bottom of the figure. Small changes in
the initial cell placement led to large
differences in duration, between storms with
short- vs. long-lived rotation. Duration here
was defined as the largest time period over which
a particular surface vorticity maximum exceeded
0.015 s-1.
15 Rotation Sensitivity
Two cases are now shown, in which the secondary
cell was southwest (left) or south-southwest
(right) of the center storm. The two cases
differ only in the 5-km difference in east-west
placement of the second cell.
30 min
2-km rainwater (shading) and surface rotation
(contoured)
16 Rotation Sensitivity
By 2.5 hours, the two simulations had diverged
significantly. The SW case had strong surface
rotation - in the hook. The SSW storm at times
had stronger - but less steady - surface rotation.
2.5 hours
SW
SSW
17 Rotation Sensitivity
Surface time series of peak vorticity from the
two cases shows the difference in steadiness.
18 Summary Conclusions
- The April, 1996 Midwest tornado outbreak has been
modeled to study convective initiation and
development. - Storm initiation occurred when dry air aloft,
which represented a modest qe gradient, became
collocated with (above) the surface drytrough in
the model fields. - Cell splitting and merging occurred in the MM5
model simulations, and was highly sensitive to
resolution. - Splitting and merging is being studied with
idealized WRF simulations of a pair of storm
cells. The evolution and mature rotational
characteristics were highly sensitive to small
changes in initial cell orientation.
19Ongoing Future Work
- Analysis of the MM5 simulation will explore the
role of the CFA in convective initiation and
propagation. - WRF simulations will take advantage of more
sophisticated model physics. In addition, future
model work will assess cell interaction within
different shear and buoyancy profiles. - Delayed cell initiation, not yet explored in our
work, will allow interaction of mature storms
with new, non-rotating ones - another mode of
interaction seen in the outbreak. - The computational demands of this work are being
used as a testbed for NCSA Portal development.
20Thank you.