Numerical simulation of the 17–18 July 1996 Chicago flood

The Chicago flood of 17-18 July 1996 was a unique weather event in Illinois because of the amount of precipitation (over 16 inches) and river flooding that occurred across the metropolitan area. Due to the proximity of a dense network of rain gauges to the heavy rainfall, this flooding event has been the subject of several journal articles discussing its meteorological and hydrological aspects. In addition, a portion of this event is the subject of a Cooperative Program for Operational Meteorology, Education and Training (COMET) flash flood and tornado forecasting case study.

Examination of observational data reveals that the Chicago flood was produced by multiple mesoscale convective systems (MCS) that developed along a quasi-stationary warm front that extended from central South Dakota to the southern tip of Lake Michigan. The convection developed and organized across western Iowa and northern Illinois first during the morning, again in the afternoon and finally in evening on 17 July 1996. The morning rainfall was produced by a decaying MCS that developed the previous evening over western Iowa. Analysis of the afternoon and evening synoptic observations reveals that the convection initiated in a region of weak upper-level divergent flow near the intersection points of a low-level jet, convective outflow boundaries and a warm front. Moderate instability and vertical wind shear across the region permitted both multicellular and tornadic supercellular thunderstorms to exist over the lifetime of this flooding event. In addition, the convective storms appear to initiate near and move along an east-west oriented outflow boundary during the afternoon resulting in heavy rainfall repeatedly crossing, or "training" over the Chicago metropolitan area. Hence, the essential elements in the Chicago event appear to be a warm frontal boundary, convective outflow boundaries, a low-level jet, and a lake-breeze from Lake Michigan. However, their role in the initiation and subsequent evolution (including their relationship to the training of convection along the frontal and outflow boundaries) are not fully understood.

Simulations of the Chicago flood of 17-18 July 1996 are being performed using the Penn. State/ National Center for Atmospheric Research (PSU/NCAR) mesoscale model (MM5). These simulations examine the ability of MM5 to reproduce the heavy rainfall event. The initial model fields are derived from analysis fields generated by the Rapid Update Cycle (RUC) numerical model. The areas of precipitation and clouds are introduced into the initial model data through subjective analysis of satellite and radar observations. Preliminary analysis of the simulations reveals that the numerical model captures several qualitative details of the flooding event including the warm frontal boundary across the Midwest and low-level jet evolution. In addition, the three-stage structure of the flood event is reproduced in the simulations with light precipitation falling in the morning hours and convective storms developing along an outflow boundary and migrating across northern Illinois in the afternoon and evening. Additional simulations are being conducted to investigate the features highlighted by previous observational studies (i.e., low-level jet orientation to atmospheric boundaries, lake-breeze) to determine their role in convective storm development and subsequent evolution.