A stochastic disaggregation method for design storm and flood synthesis

Abstract A simple technique for short-scale rainfall disaggregation is developed and studied both theoretically and empirically. This technique can be combined with a variety of rainfall models. The simplest implementation of the technique for a Markovian structure at a discrete time with only three parameters is studied in detail as an easy and convenient engineering tool for design storm and flood studies. Combining the disaggregation technique with a succession of simple hydrological models, i.e. a production function, a unit hydrograph and a flood routing model, we form a stochastic approach for design storm and flood synthesis. Similar to common engineering methods, the proposed method starts with the selection of certain characteristics of the design storm (i.e. its duration and total depth that correspond to a given return period). Subsequently, the method generates a series of probable time distributions by disaggregating the given total depth into incremental depths. Then the series of hyetographs is routed through the hydrological models and the result is the conditional probability distribution function of the outflow peak of the hydraulic construction studied, given the duration and total storm depth. From this distribution we can adopt the design discharge either as the conditional expected value of the outflow peak or as a value corresponding to a selected probability level. The method is illustrated with a real-world example and compared with common engineering methods.