The following web page, created by Albert Pietrycha, is an expanded version of the web page that resides on the Chicago National Weather Service web site. The expanded page below consists of links to tornado pictures, satellite and radar animations, and expanded commentary. For an enhanced study investigating various aspects of the event, please read the following paper submitted to the AMS 2004 Severe Storms Conference; a pdf file. To view related aspects to the paper, see Jon Davies great web page.
|1. Event overview
On the late afternoon and early evening of 20 April 2004, an outbreak of tornadoes occurred across northern Illinois and central Indiana, including
an F3 tornado that killed eight people. Over a four and a half hour period, 30 tornadoes occurred from discrete ‘low-topped’ supercells that developed along
a rapidly advancing warm front. The storms formed in an environment where convective available potential energy (CAPE) was relatively small compared
with typical tornadic supercells. Several of these supercells produced more than five tornadoes each. Some of the tornadoes could be classified as large
long tracked (e.g., 0.8 km wide with path lengths greater than 24 km).
Earlier in the day widespread rain and dense stratiform cloud cover existed across northern Illinois, north of a warm front that lay draped across southern Illinois. South of the front ample moisture and clearing skies existed. By mid afternoon the front began to move rapidly northward. With the clearing skies and improved low-level moisture, atmospheric instability increased quickly (50 mb mixed CAPE near 1100 J kg-1 ) across northern Illinois. The instability combined with strong winds aloft provided conditions favorable for supercell thunderstorm development. The storms first initiated over west central Illinois then moved northeast near 30 mph. Click here for a visibile satellite animation with 850 mb profiler winds. The 7.3 meg file should open in your brower.
A break down of the supercell storms are as followings.
Storm A: Traveled from Marshall to LaSalle Counties - produced seven tornadoes.
An amended Public Information Statement below contains additional information about the tornadoes.
PUBLIC INFORMATION STATEMENT
...APRIL 20 TORNADO UPDATE...
AFTER THREE DAYS OF EXTENSIVE GROUND AND AERIAL SURVEYS...AND
IN EASTERN STARK COUNTY AND WESTERN MARSHALL COUNTY AN
IN SOUTHERN BUREAU COUNTY THE SECOND TORNADO TRAVELED NEAR
THE THIRD TORNADO DEVELOPED ABOUT 2 MILES WEST SOUTHWEST
A FOURTH TORNADO FORMED ABOUT 2 1/2 MILES EAST NORTHEAST
THE SIXTH TORNADO WAS 1 MILE LONG AND 30 YARDS WIDE. THE
THE SEVENTH AND FINAL TORNADO FROM THIS STORM OCCURRED FROM
ANOTHER STORM PRODUCED TWO TORNADOES. A TORNADO OCCURRED
AFTER THIS STORM MERGED WITH ANOTHER STORM OVER WILL COUNTY...IT
ANOTHER STORM PRODUCED A SERIES OF FIVE TORNADOES OVER
A SECOND TORNADO OCCURRED 10 MILES WEST OF ASHKUM ON THE
A THIRD TORNADO FROM THIS STORM DEVELOPED 7 MILES SOUTHWEST
ANOTHER TORNADO OCCURRED FROM 1 MILE SOUTHWEST OF GRANT
PARK TO 2
THE FIFTH AND FINAL TORNADO FROM THIS STORM OCCURRED 3
ANOTHER STORM PRODUCED A TORNADO ON THE NORTHWEST CORNER
The following are links to various websites containing pictures of the Utica tornado. Please respect their copyrights.
|2. Damage survey maps
Immediately following the event five teams conducted ground based and aerial damage surveys across the Chicago NWS CWA. The map to the right is a broad overview of the tornadoes that occurred within, or very near the Chicago CWA.
||Fig. 1. Tornado paths (bold blue line) and damage ratings are depicted. Towns in bold type are reference locations as mentioned within the Public Information Statement.|
|The map to the right is of the tornado that struck both Granville and Utica. The tornado traveled from southwest to northeast. Note how the tornado crossed the Illinois River two times. I mention this a there appears to be many folks out there who believe the myth that tornadoes don't cross rivers!||
||Fig. 2. As in figure 1.|
|The map to the right is of the tornado path that struck Utica. The tornado traveled from southwest to northeast.||
||Fig. 3. Progressive gray shading depicting areas of F0 to F3 damage. The survey is to scale.|
|The map to the right is of the tornado that struck Joliet. The tornado traveled from southwest to northeast.||
||Fig. 4. The light gray shading respesents the swath of F0 damage. The darker gray depicts areas of F1 damage.|
The following photos of swirl marks and structural damage were taken on 23 April 2004. As time permits commentary will be provided.
||Fig. 6. A couple of lucky barns southwest of Kankakee. View is to the SSW from ~1500 ft agl.|
3. Meteorological data
a. Upper air
A deep, negatively tilted, upper-level trough was present over the central and northern plains at 1200 UTC 20 April 2004. The 1200 UTC 500 mb analysis indicated a 50 knot speed maxima rounding the base of the upper trough, spreading from the southern Rockies into southern Kansas and Oklahoma, while farther to the northeast, an upper jet speed maxima of 110 knots was analyzed at 250 mb over the upper Mississippi valley and western Great Lakes. Strong low-level warm air advection was occurring over northern Missouri and Iowa on the nose of a 40 knot 850 mb jet, under an area of strong upper level divergence in the right entrance region of the upper level jet streak. The upper trough continued to deepen during the day as it translated into the middle and upper Mississippi valley.
||Upper air charts use a standard station model in abbreviated format; temperature and dewpoint in Centigrade. Winds as in the surface maps.|
Partial analysis and discussion was provided in the previously mentioned SLS paper (see the link above) concerning the RUC and Davenport (DVN) soundings . However, the paper does not provide information on the Lincoln (ILX) soundings, nor the 12Z DVN sounding. To see the ILX soundings click here. For the DVN soundings click here.
||CAPE/CIN calculations were derived from lifting the lowest 50 mb mixed layer and using a virtual temperature correction. .|
In response to the UA features described above, surface low pressure developed over the plains, with the 1200 UTC surface analysis (not shown) showing a 1003 mb low over Nebraska. East of the developing low, a warm front stretched across northern Missouri and southern Illinois into the lower Ohio valley. South of the warm front, dewpoint temperatures ranged 15.5-18.3ºC, while north of the front, in an area of extensive stratiform cloud cover and rain showers, dewpoints ranged 5-9ºC.
Over the course of the late morning and afternoon the surface low deepened to 996 mb and became located over northern Iowa by 0000 UTC 21 April 2004. Across the region of interest, a quasi-stationary front remained draped across the Illinois / Wisconsin border. The stationary front began to mix slowly north by late afternoon (depicted in figure 5as the "northern most warm front"). The main (southern) warm front mixed north near 30 kts and became established across the Chiago CWA by 2100 UTC. The arrival was some 2 to 3 hrs faster than progged progged by the ETA and the RUC-40. The two models mixed the warm front through the Chicago NWS CWA between 23 to 03 UTC that evening, a time period when the contribution to instability enhancement by solar heating would become minimized.
The following links are to unanalyzed surface maps. All times UTC.
||Fig. 5. Surface map for 00 UTC 20 April 2004 . Standard station model used; temperature and dewpoint (°F), pressure (mb), 3-h pressure falls, and sky conditions reported. Winds in knots with one full barb, and one half barb equal to 10, and 5 knots, respectively. Warm fronts depicted with conventional symbols. Black arrows indicate the direction of forward frontal propagation. Green arrow represents axis of largest surface theta-e advection. Purple shading denotes area of 3-h pressure falls > 4 mb 3 h-1.|
A radar animation of the KLOT 0.5 degree reflectivity data can be downloaded. The 10 meg file should open within your browser. The animation spans five hours, beginning with stratiform rain across the region followed by a brief break prior to the development of the storms. The mean motion of the activity was to the northeast near 30 kts. Note the left moving storm into the forward flank of storm C immediately prior to the storm producing the Joliet tornado. Then, poof... storm C rapidly dissipates.
The following National Weather Service radar imagery is near the time the town of Utica was struck by a tornado (Storm A). Both the 8-bit 0.5 degree reflectivity and storm-relative velocity data are shown. The storm was 45 nm away from the radar with a center beam height of ~ 3900 ft above radar level.
||Fig. 7. Radar reflectivity near Utica at 6:09 PM (2309 UTC), 20 April 2004. The warmer colors to white represent regions of greatest rainfall and hail within the storm.|
||Fig. 8. As in figure 7 but depicting storm-relative velocity data. The cool colors represent motion toward the radar. The warm colors represent motion away from the radar.|
|The National Weather Service radar imagery to the right is at the time of the Joliet tornado (Storm B). Both the 8-bit 0.5 degree reflectivity and storm-relative velocity data are shown. The hook echo and mesocyclone range from 4-9 nm away from the radar with a center beam height of ~ 300 ft above radar level.||
||Fig. 9 As in figure 9 but for 7:44 PM, 20 April 2004.|
||Fig. 10. As in figure 8 but for 7:44 PM, 20 April 2004.|
The storm relative velocity image (Fig. 11) depicts several noteworthy features. I should add there was strong continuity in the neighboring elevation slices to the following features. Unfortunately KLOT level-II data are not available for this event which prevents the ability to generate radar cross sections. The data appear to have resolved a portion of the rear flank downdraft (RFD) gust front and a portion of the occlusion process. At such close distance to the radar (< 9 nm), the circulation associated with the mesocyclone is clearly evident. Not surprisingly, the tornado cyclone and occlusion downdraft were to small in horizontal scale to be resolved by the radar. Also of note is the enhanced region of inflow. There is some question whether the velocity signature toward the rearward edge of the RFD gust front is related to a downburst, or random chaotic motion.
A 4-frame Java animation of storm relative velocity is available to view. The imagery is at the 3.4 degree elevation slice.
||Fig. 11. A 3.4 degree storm relative velocity image depicting the RFD at 7:40 pm CDT.|
|The National Weather Service radar imagery to the right is at the time of the Kankakee tornado (Storm D). Both the 8-bit 0.5 degree reflectivity and storm-relative velocity data are shown. The storm was 31 nm away from the radar with a center beam height of ~ 2400 ft above radar level. The storm straddling the Illinois-Indiana state line produced the tornado in Hopkins Park (Storm E).||
||Fig. 12. As in figure 7 but for 7:39 PM, 20 April 2004.|
||Fig. 13. As in figure 8 but for 7:39 PM, 20 April 2004.|
|The image to the right is of the Chicago National Weather Service Radar, wind profile from 4:53 to 5:44 pm (2153 – 2244 UTC). Note the increasing, nearly unidirectional flow and deepening mid-level jet. This type of flow régime is common with some supercell events.||
||Fig. 14. KLOT WSR-88D Wind profile. Winds as in figure 5.|
A big thank you to WBBM CBS Channel 2 for providing aerial damage survey assistance and to Jim Allsopp at the Chicago NWS WFO for the survey photos.
Return to Albert Pietrycha's homepage.