No Do-Overs for Plainfield, Please

If ever there was a day meteorologists might like to do over, it was exactly 20 years ago today, on 28 August 1990. Somehow, on an afternoon originally projected to have a mere moderate risk of severe weather, an F-5 tornado—the only such powerful tornado in August in U.S. history—struck northern Illinois, killing 29 people in the small town of Plainfield, just 30 miles southwest of Chicago.
The sky turned black, and few people knew what happened as the rain-wrapped tornado ripped through the landscape. Almost no one saw the funnel. (Paul Sirvatka, just then starting up his College of DuPage storm chasing program was a rare exception.) Even though another tornado was spotted earlier in the afternoon in northern Illinois, no sirens wailed in Plainfield until too late. No tornado warnings were issued until too late.
Tom Skilling (WGNE-TV) broadcast a report this week for the 20th anniversary of the tragic tornado, explaining why warnings would likely be much better should similar weather now threaten the Chicago area.

The gist of the story: back in 1990 Chicagoland didn’t have NEXRAD Doppler radar and other recent advances in observing and modeling. Also, the aftermath led to a reorganization of the overworked Weather Service meteorologists in Illinois, narrowing the purview of the Chicago office and adding more offices to help cover the state.
While most stories in the media (for example, also here) have been showing why 20 years have made a repeat of Plainfield’s helplessness less likely, Gino Izzo of the NWS Chicago office decided to have a do-over anyway–on the computer. At his presentation for the AMS Conference on Broadcast Meteorology in June, Izzo described how he reran the severe weather forecasts for 28 August 1990 using the North American Regional Reanalysis and the most up-to-date model of 2010, the Weather Research & Forecast model (WRF) from UCAR.
With a nested 4 km grid at its most detailed, Izzo ran the WRF overnight–it took 10 hours on the available computer in the office–and found that in fact, with the observational limits of 1990, the latest, greatest numerical forecasting doesn’t really improve the severe weather outlooks for the Plainfield disaster. The WRF moved the likely areas of severe weather (not tornadoes necessarily, but probably winds associated with a bow echo) too far eastward. Only when the model time horizon was getting within a few hours of the killer tornado did the severe weather outlook for northern Illinois start to look moderate, as the model begins to slow down the eastward progress of the cold front.
Audio and slides from Izzo’s striking presentation are available on the AMS meeting archive. The message is pretty clear: no matter how good the models get, Doppler radar, wind profilers, aircraft-based soundings, and satellites make a huge difference in our severe weather safety these days.
Of course, with or without better warnings, a repeat of the Plainfield disaster would be potentially catastrophic. The area has more than doubled its population since 1990. And 28 August just happened to be one day before school resumed for the fall—few people were at the high school that was totally destroyed that day. Even just a day’s difference, let alone two decades, could have been critical. Nobody wants a do-over.