Improving conceptual hydrological modeling under uncertainty
La Trobe Event Space (Entrance at 10 La Trobe St.)
Building 381 Ground Floor
333 Exhibition Street, Melbourne 3000
PhD completion seminar by Sina Khatami
Hydrological models are conventionally evaluated in terms of their response surface or likelihood surface constructed with the model parameter space, based upon the concept of the equifinality of model structure/parameters. Examining various aspects of this convention, I demonstrate their shortcomings and introduced improvements. First, arguing that there is more to equifinality than just model structure/parameters, I propose a theoretical framework for equifinality for scientific inquiry and hypothesis testing under uncertainty. Second, I developed a new method for model evaluation — called flux mapping — based on the equifinality of model internal fluxes. Evaluating the model response within the flux space is a powerful approach, beyond the convention, to evaluate model process representation and formulate testable hypotheses. Third, I further demonstrate to what extent flux maps are dependent upon the choice of error metric, data information content, and adequacy of parameter sampling. I particularly demonstrate the fundamental shortcomings of two widely used error metrics — i.e. Nash–Sutcliffe efficiency and Willmott’s index of agreement — for model evaluation, and argue the Kling–Gupta efficiency is a more reliable error metric, while it is not without its own flaws. Finally, I classify a set of Australian catchments (known as HRS catchments) based on their hydrological similarity in terms of their corresponding flux maps.
Sina Khatami , PhD candidate