Temporal and Spatial Characteristics of Heavy-Precipitation Events in the Midwest

Abstract

Long-term (1921?90) daily precipitation data from 242 stations were used to identify heavy multiday precipitation events (exceeding the threshold for a 1-yr recurrence interval) that were found to be closely related to flood events. The number of events were aggregated over 5-yr (pentad) periods and compared with total pentad precipitation. Although a strong positive correlation was found, this was due entirely to the event contributions to the total precipitation. When event precipitation was subtracted from total precipitation, no statistically significant correlation was found. The frequency distribution of precipitation totals in nonevent weeks was also found to be similar in years with few or no events compared to years with several events. These findings suggest that the occurrence of these events is not strongly linked to longer-term persistent climate anomalies. These relatively few events make an important contribution to long-term precipitation variability, accounting for about half of the interpentadal variability. To provide information for determining hydrologic impacts from the results of GCMs, a study of spatial precipitation variability during heavy events was undertaken. For each event at each station, grid-average precipitation was calculated for 2° latitude ? 2.5° longitude, and 4° latitude ? 5° longitude grid cells. The ratio of grid average to heavy-event precipitation totals was determined. These relationships could be used to assess the probability of flood-producing, localized precipitation extremes from GCM grid-average precipitation.