A conceptual systems model of rainfall-runoff on the Haast River

A conceptual systems model of runoff generation is described and applied to hourly rainfall-runoff data from the Haast River. The so-called MaxH model treats the catchment as a statistical population of different-sized water storage elements. The distribution of water within this population is also described statistically. The uncertainty over the choice of probability distributions for the catchment storages and the actual catchment storage within them is dealt with using the principle of maximum statistical entropy. The characteristics of the derived nonlinear model are described and compared to the more familiar linear reservoir. Both model types are then applied to a set of rainfall-runoff data from the Haast River in the western Southern Alps on the South Island of New Zealand. Within the limitations of the study, the MaxH model displays some useful characteristics such as the ability to reproduce a more rapidly rising hydrograph limb with increasing catchment wetness. In other respects, some of the long-standing problems with simple catchment models still remain, such as finding objective ways of determining the percentage runoff, baseflow separation etc.