Received: 12 May 2016 – Discussion started: 18 May 2016
Abstract. Design flood, river discharge with a particular return period, is fundamental to determine the scale of flood control facilities. In addition, considering a changing climate, not only frequencies of river discharge at design level but those of devastating flooding are also crucial. Characteristics of river discharge during extreme floods largely differ from those during others because of upstream dam operation and/or river overflow; however, flood frequency analysis (FFA) from past discharge data is difficult to represent such impact because river basins rarely experience floods over the design level after river improvement and dam construction. To account for the above impact on extreme flood frequencies, this study presented a rainfall-based flood frequency model (RFFM) that derives flood frequencies from probabilistic rainfall modelling that empirically represents probabilistic structure of rainfall intensity over a catchment by directly using observed spatial-temporal rainfall profiles. The RFFM was applied to the Yodo River basin, Japan and demonstrated that flood frequency estimations by the RFFM well represent past flood frequencies at the Hirakata gauging station. Furthermore, the RFFM showed that return periods of large flood peaks are estimated at extremely large values, reflecting decrease of discharge by the inundation in an upstream area of the gauging station. On the other hand, FFA from past discharge data did not represent this impact because it has not experienced such huge flood peaks in an observation period. This study demonstrated the importance of the RFFM for flood frequency estimations, including those exceeding the design level.
How to cite. Tanaka, T., Tachikawa, Y., Ichikawa, Y., and Yorozu, K.: Frequency analysis of extreme floods in a highly developed river basin, Hydrol. Earth Syst. Sci. Discuss. [preprint], https://doi.org/10.5194/hess-2016-225, 2016.
This study proposed a new flood frequency (FF) model, which derives FF from probability distribution of rainfall. It showed good agreement with flood frequency analysis of past discharge data in a large river basin (7290 km2). Furthermore, by introducing an inundation model to an upstream area, the proposed method represented quite small frequencies of extreme floods, which cannot be explained by FF analysis. The proposed FF model is useful for (extreme) flood frequency estimations.
This study proposed a new flood frequency (FF) model, which derives FF from probability...