Common Unitgraphs for Sets of Runoff Events. Part 2: Comparisons and Inferences for Rainfall Loss Models

Abstract

Common unitgraphs obtained from streamflow data for a set of runoff events have higher peaks, shorter times to peak and generally shorter total durations than average unitgraphs obtained by conventional techniques from rainfall and streamflow data for single events. The calculated inputs to these unitgraphs lag behind the corresponding peaks in the catchment rainfall and continue after the end of rainfall. It eventuates that these inputs are best modelled by including a strongly non-linear storage in the rainfall loss model. For common loss and storage parameters, runoff hydrographs from the overall model fit the observed hydrographs over a wider range of flows than predictions from application of conventional unitgraph methods. The new approach can be interpreted physically as a non-linear process of rainfall loss and routing to stream channels, followed by linear routing of stream flow; or functionally as a separation of the common linear elements in the overall rainfall-runoff process from the loss and non-linear elements. It has the potential to lead to improved predictions of storm runoff hydrographs.