SIMILARITY ANALYSIS OF RUNOFF GENERATION PROCESSES IN REAL-WORLD CATCHMENTS

This paper addresses the question of similarity of runoff generation processes between catchments in the eastern wheat belt of Western Australia, and the use of dimensionless parameterizations to quantify this similarity. A spatially distributed rainfall-runoff model, simulating runoff generation by both the infiltration excess (Horton type) and saturation excess (Dunne type) mechanisms, was developed for catchments in the region. Seven small experimental catchments, with field-measured soil hydraulic properties and topography, were used in the study. Following on from the similarity theory developed by Sivapalan et al. (1987), a number of dimensionless similarity parameters were constructed using the field-measured soil and topographic properties, a characteristic length scale, and a characteristic flow velocity. The objective was to determine whether the dominant runoff generation mechanism on a catchment could be reliably predicted by these similarity parameters. This was achieved through sensitivity analyses carried out with the rainfall-runoff model. Two dimensionless parameters, K0* and f*, were found to be critical for characterizing the similarity or dissimilarity of the runoff generation responses between the seven experimental catchments. Within the assumptions of the analysis, two catchments in the wheat belt region can be considered to be hydrologically similar, in terms of their runoff responses, if K0* and f* are identical in both catchments. The dominant mechanism of runoff generation on any catchment can be reliably predicted, provided that the values of K0* and f* are known. A partial quantification of the Dunne diagram (Dunne, 1978) for the wheat belt region, in terms of the infiltration excess and saturation excess mechanisms, was achieved by artificially varying K0* and f* in the rainfall-runoff model to explore the full range of possible runoff generation responses.