Proceedings, 19th Conference on Severe Local Storms,
American Meteorological Society - Minneapolis, Sept., 1998



Brian F. Jewetta, Bruce D. Leeb and Robert B. Wilhelmsona

aDepartment of Atmospheric Sciences and
National Center for Supercomputing Applications (NCSA)
University of Illinois at Urbana-Champaign
bDept. of Earth Sciences, University of Northern Colorado


On April 19, 1996, an outbreak of tornadic storms struck parts of Illinois, Iowa, Missouri, Indiana and Kentucky. The single-day total of 30 tornadoes was the highest ever reported in Illinois, where 31 counties sustained damage. While most tornadoes were short-lived, some reached F2-F3 intensity, with over $30 million in damage and one fatality in Illinois (Storm Data, 4/96). Beneath the left exit region of a 65 m s-1 upper-level jet streak, a deepening surface cyclone moved from southern Iowa into northwestern Illinois. The intense (992 mb) cyclone was accompanied by surface dewpoints over 18°C, strong southeasterly surface winds and substantial low-level shear. Thunderstorm development began by early afternoon, with the first severe thunderstorm warning issued near 2100 UTC.

Severe thunderstorms occurred along both the surface warm front, which extended from southern Iowa into eastern Illinois, and along a north-south surface pressure trough/dryline which moved into Illinois from Missouri. Preliminary analysis of WSR-88D data indicates that the storms along the dryline underwent a pattern of cell splitting and merging prior to intensifying into a broken line of tornadic storms that moved across Illinois.


We are modeling the April 19 cyclone and squall line with the NCAR/Penn. State MM5 model (Grell et al., 1995). Sixty-six vertical layers are employed (up to 60 m resolution), with horizontal resolution from 81 km (outer grid) to 1 km (planned). Two-grid results (27 km resolution) to date show promise (Fig. 1)

Fig.1. Left: 0000 UTC 20 April 1996 radar reflectivity (only values exceeding 15 dBZ shown).
Right: 24-hour simulated low-level rainwater field from MM5, valid at the same time.
(color radar figure available here - not in preprint)
but important differences from observations remain in both the mature squall line structure (even while allowing for the coarse resolution) and in details of the mesoscale environment near the time of convective initiation. Simulations with 9 km resolution were similar, so improvements in the model initiation and/or boundary tendencies are necessary. Data assimilation experiments are underway with an emphasis on a more accurate representation of the dryline by mid-late afternoon on April 19th.

Our goals in this case include understanding the thunderstorm cell-cell interaction which occurred in the hour prior to the first tornado touchdown in western Illinois, and determining the extent to which this evolution was important in the development of long-lived tornadic thunderstorms on this day. We will also use the MM5 model fields to extract vertical soundings along the prominent mesoscale boundaries (warm front and dryline). These soundings will be used to perform thunderstorm simulations in idealized isolated (horizontally uniform) conditions.

The differences between the simulated convection under complete vs. homogeneous environments should help reveal the importance of the surface boundaries in the actual thunderstorm evolution.


Simulations are being carried out on the NCSA CRAY Origin2000. Additional computing and other support is also provided by NCSA. This work is supported by the National Science Foundation under grant ATM 96-33228.


Grell, G. A., J. Dudhia, and D. R. Stauffer,
1995: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Technical Note TN-398+STR, 122 pp.

National Climatic Data Center, 1997: Storm
Data, 38, No. 4 (4/96).

Brian F. Jewett | Convective modeling group | publications