A quantile regression technique to estimate design floods for ungauged catchments in south-east Australia

Abstract

This paper develops prediction equations for design flood estimation in small to medium-sized rural ungauged catchments in south-east Australia using a quantile regression technique. These cover six average recurrence intervals in the range 2 to 100 years and require five climatic and catchment characteristics variables as inputs. An independent test indicates that the developed prediction equations can provide reasonably accurate design flood estimates for a range of applications. The method does not require assumption of geographical proximity as with the runoff coefficients in the Probabilistic Rational Method and can provide design flood estimates that are more accurate than the Probabilistic Rational Method.