FLOOD RESPONSE TO TROPICAL AND EXTRA-TROPICAL

STORM SYSTEMS ON THE CAROLINA COASTAL PLAIN

Laura T. Neal

Frequency analysis of streamflow assumes that flood magnitude can be treated as a random variable drawn from a population with a known probability distribution (log normal, Gumbel, log Pearson, etc). This assumption may be false in watersheds where floods are generated by distinct hydrologic processes creating mixed statistical populations. For example, many humid subtropical watersheds in the southeastern United States have flood frequency regimes that are the product of separate populations of precipitation events that dominate in summer versus winter.

This study demonstrates significant differences between floods generated by tropical and extra-tropical storm systems for 4 medium-sized rivers (1,700-3,200 sq. km.) in North and South Carolina. Late summer and early autumn floods on these rivers are generated by intense convective precipitation from tropical depressions, storms, and hurricanes. Winter and spring floods are generated by widespread precipitation associated with passing frontal systems and midlatitude cyclones. Winter floods occur 2-3 times more frequently than summer floods in the annual flood series of these rivers, but most of the largest floods on record were generated by tropical systems. Summer floods are generated by precipitation of greater intensity, but are preceded by warm weather and lower baseflow conditions. Winter floods are associated with greater average peak discharge, reflecting the moderate intensity and large spatial coverage of frontal storm systems. The frequency distributions for summer floods show significantly lower peak discharges, with much greater standard deviation, skewness, and kurtosis than the winter flood distributions. Therefore, although summer floods are less frequent than winter floods, they strongly influence the parameters used in flood frequency prediction methods.