Calibration procedures for rational and USSCS design flood methods

Abstract

Flood estimation models are often used to predict design floods of a selected average recurrence interval (ARI) using rainfall information of the same ARI. That is, they are used for probabilistic purposes rather than for forecasting during actual events. Design values of model parameters need to be derived in a manner corresponding with the way the models are applied in design. Design values of the main parameter of two simple flood estimation models, the rational method (runoff coefficient) and the United States Soil Conservation Service (USSCS) method (curve number), were derived for 105 small agricultural catchments in and around south east Queensland, Australia. These were obtained by reversing the way the models are used in design flood estimation, utilizing frequency analyses of rainfall and flood data. The derived values were considerably different from conventional handbook values; the latter gave inaccurate estimates of design runoff and flood peaks. The values derived from observed data were only weakly related to catchment characteristics including land use and soil type, but were regionally consistent. ARI and the method of estimating design rainfall duration had greater influences on the values derived.