A methodology for hypothesis testing in conceptual catchment modelling

During the latter part of the 1980s there was some debate in the literature that hydrology was not practised as a science but as an engineering technology. Hydrologic modelling was seen as devoid of a scientific basis, conceptual catchment modelling in particular, because the practice of calibrating a model to the response that was the subject of investigation was seen as “fudging” a catchment water balance. Presented here is a methodology, based on Popper’s (1959) theory of the nature of the scientific method, to more rigorously test the hypotheses about the catchment water balance that constitute a conceptual catchment model. The methodology has three main elements. Firstly, the calibration of a conceptual catchment model to multiple catchment responses to constrain the model behaviour. Secondly, the application of established statistical techniques to critically assess model predictions and errors; and, thirdly the principle of parsimony, the search for the simplest model that fully explains the data. The methodology is demonstrated by way of a case study in which a conceptual catchment model is proposed and rigorously tested. This is followed by a comparison with another model developed from the same conceptual base, but without reference to the proposed methodology.