Development and Assessment of a Bimodal Synthetic Design Storm Using Observed Data

The efficiency and robustness of urban stormwater infrastructure requires review, particularly regarding devastating double-peak (bimodal) storms that have been observed in various regions globally but are not adequately represented by current synthetic design storm methods. This study investigated the occurrence and characteristics of bimodal rainfall events in a case study undertaken in eThekwini, South Africa, and developed a representative synthetic bimodal design storm for urban hydrological modelling applications. A novel, objective methodology was developed and implemented to isolate bimodal events from observed rainfall data. Analysis revealed that twelve percent of major storms exhibited a bimodal distribution, which were typically more extreme and of longer duration than single-peak events. Two new triangular design storms, Bimodal-Same Recurrence Interval and Bimodal-Split Rainfall, were developed based on the observed bimodal rainfall characteristics. Comparative evaluation showed that the Bimodal-Split Rainfall method offered the most reliable overall representation of bimodal events. This method accurately reproduced total storm volume and effectively captured the timing of peaks and saddle points. The Bimodal-Split Rainfall approach provides a meaningful improvement over conventional single-peaked design storms and is recommended for adoption in local stormwater design guidelines in eThekwini. The methodology developed can be used to generate bimodal design storms in other regions where bimodal storms occur.