The 1995 Chicago heat wave: How likely is a recurrence?

The deadly heat wave of July 1995 that affected much of the U.S. midwest, most notably Chicago, Illinois, has been put into historical perspective. The heat wave has been found to be remarkably unusual, but only partially because of the extreme high apparent temperatures (an index of the combined effect of temperature and humidity on humans), when the authors calculate a return period of the peak apparent temperature of less than or equal to 23 yr. Of greater significance were the very high temperatures that persisted day and night over an extended 48-h period. Analysis presented here indicates that for Chicago such an extended period of continuously high day and night apparent temperature is unprecedented in modem times. The 2-day period where the minimum apparent temperature failed to go below 31.5 degrees C (89 degrees F) is calculated to be an extremely rare event (probability of occurrence < 0.1%) based on a 10 000-yr-long simulation of a four-parameter (temperatures related to the mean, the intraseasonal daily variance, the interannual variance, and the day-to-day persistence of temperature) probabilistic model. Such unusual heat waves evoke questions related to the future course of the climate and whether this recent event was merely an extreme anomaly or part of an ongoing trend toward more extreme heat waves. A Monte Carlo analysis of trends (1948-95) for various quantiles of the hourly apparent temperatures during the most severe heat waves each year from 26 midwestern stations reveals a modest, statistically insignificant increase of apparent temperatures for a wide range of quantiles without the inclusion of 1995 data. There is a statistically significant increase in apparent temperature with its inclusion, reflected most strongly for upper quantiles or daytime temperatures. It is argued, however, that because of the impact of changes in instrumentation at primary National Weather Service stations, the potential affects of urbanization, and little trend of summer mean temperatures, it is unlikely that the macroscale climate of heat waves in the Midwest or in Chicago is changing in any significant manner. Trends notwithstanding, the authors demonstrate the difficulty associated with projecting changes in the frequency and severity of similar types of events, even if the mean apparent temperature could be accurately predicted for the next century, for example, global warming projections. This is demonstrated using Chicago temperatures. The authors show that accurate projections of the frequency, severity, and duration of heat waves in the Midwest require accurate projections not only of the mean, the interannual variance, the intraseasonal variance, and day-to-day persistence, but also the interrelationships among these quantities within different synoptic-climatic regimes.