Development and Application of a Large Flood RegionalisationModel for Australia

Estimation of large to rare floods is imperative in planning and designing large hydraulic structures. Due to the limited availability of observed flood data, the estimation of large to rare flood frequencies requires significant extrapolation beyond the available flood and rainfall data. Notable methods for estimation of large to rare floods include rainfall-runoff models and empirical approaches. This paper presents the further development of a simple Large Flood Regionalisation Model (LFRM) based on observed flood data which in practice will be relatively easier to apply. The LFRM assumes that the maximum observed flood data over a large number of sites in a region can be pooled together by accounting for the at-site variations in the mean and coefficient of variation (CV). The LFRM builds on previous research by adding a spatial dependence model which accounts for the net information available for regional analysis (i.e. number of independent sites Nₑ). The LFRM can be been applied to ungauged Australian catchments for the estimation of floods with AEPs up to 1 in 3000. It shows little bias and compares very well to the World model