EXTREMES FOR RAINFALL AND STREAMFLOW, HOW STRONG ARE THE LINKS?

Abstract

Many hydraulic structures are designed using calculation procedures that assume an equivalence between the return period of a rainfall event and the return period of the streamflow peak calculated using that rainfall. In this study, we have examined both observed peaks of streamflow and the meteorological sequence creating the peak to check the validity of the assumed relationship between rain and peak flowrate. We used observations from five rural watersheds in southern Ontario with areas ranging from 2 to 150 km2. In these watersheds we found that few annual flood peaks were created by rain or by snowmelt alone; 80% of them were combined events. Ten percent of those events in the annual series with return periods greater than two years were caused by rain alone. Very few of the annual extreme rain events, defined for a duration set at the "time of concentration" of each watershed, caused flow peaks that appeared in either the annual flood series or the annual rain-induced flood series. The weakness or lack of causal link between extreme rainfall and extreme flows on rural watersheds demonstrates the need for research into methods of calculating flood peaks that do not assume such linkage.