Estimating design floods based on the critical storm duration for small watersheds

Abstract The objective of this study was to propose a new method to determine design floods using the critical storm duration concept. Five different models, including the Rational, SCS, and Clark methods, were used to estimate peak discharges, while the uniform distribution, Mononobe, Huff, and Yen and Chow methods were applied for the determination of temporal rainfall distribution. Two small watersheds, Baran (HP#6) and Banweol (WS#1), for which watershed hydrologic data were available since 1996, were selected as the study areas. A total of 41 rainfall events was chosen from the study watersheds to calculate peak runoffs and evaluate the performances of the selected hydrological models based on the statistics of RMSE, Nash efficiency criterion (NEC), and R2 value. The Clark method performed the best overall among the selected models, with both NEC and R2 values greater than 0.95. The Huff method resulted in the longest critical storm duration, which was much greater than the times of concentration. The increase in the recurrence interval decreased the critical storm duration while increasing the peak flow rates. The SCS model estimated the greatest design floods, 94.2 m3/s for HP#6 and 56.4 m3/s for WS#1, with a 25-year return period. The design floods for the study watersheds that were estimated by the selected hydrologic models ranged from 68.3 to 132.1% of those estimated by the Rational method. The greatest to the smallest peak flows resulted from the SCS, WFRP, Clark and Nakayasu methods in order. As an alternative to the Rational method, the WFRP method may be appropriate for rural watershed areas in Korea, where paddy fields commonly exist, whereas the SCS model may be more suitable for urban areas, where most land surfaces are covered with impervious material. It was concluded that the incorporation of a critical storm duration concept can contribute to the advance of design flood estimation method in Korea.