Threshold analysis of rainstorm depth and duration statistics at Toronto, Canada

Storm-event based probabilistic models characterizing the probability distributions of storm depth and duration are developed. They overcome the primary limitation of conventional rainfall depth–duration–frequency (DDF) analysis. The application of threshold-excess extreme value analysis techniques to storm-event statistics provides a simple, statistically efficient means of characterizing frequency of extreme storm-event depths and durations. The best-fit probability density functions of storm-event depth and duration are used to derive storm-event depth–duration–frequency combinations. The resulting models may be used to develop design storms based upon actual storm-event statistics. Comparisons between conventional and storm-event analysis (SEA) models highlights the improvements and benefits of using storm-event-based probability distributions, and brings into question the application of conventional DDF techniques for design storm development.